Compass Meridian, Mississippi

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

Compass offers a private wholesale-development route into Mississippi compute demand. A long-term utility agreement and reporting of work on the first two buildings support construction-stage coverage. These milestones do not establish occupied capacity or investment returns.

Site and development scope

The campus is in Meridian’s I-20/59 Industrial Park, Lauderdale County. Plans describe eight single-tenant buildings developed over eight years. Use a locality marker near Meridian rather than an exact parcel centroid.

Power and capacity

Mississippi Power says it signed a nearly 500 MW long-term agreement. Separately, Compass told DCD each planned building would provide 40 MW of IT capacity. Utility service capacity and IT capacity are different measures; neither is verified operating load.

Capital and financing

The announced $10 billion investment includes future tenants’ IT equipment and full campus fitout. It is not Compass-only construction spending. Project-specific financing commitments, loan closure, lease economics and expected returns were not established.

Demand and customer evidence

MDA announced Corderill LLC as a tenant on April 22, 2026, with at least $100 million planned equipment investment and 20 jobs. This supports an identified tenant role, but does not establish computing delivery, an ultimate hyperscaler customer or customer equipment installed.

Company economic roles

Compass develops and operates single-tenant campuses; Corderill is an announced tenant and equipment investor. Mississippi Power provides utility service. State and local bodies support incentives and site infrastructure. No project-specific GPU or other vendor contracts were established.

Milestones and readiness

The project was announced January 9, 2025; WTOK covered the January 28 groundbreaking. July 21, 2026 reporting described work on the first two buildings and a first-building target near year-end 2026. Classify construction underway; the target is not completion or occupancy evidence.

Government evidence

MDA’s April 22, 2026 record says Corderill’s incentive was approved based on expected employment. Incentive certification does not establish construction completion. Campus building and environmental permit identifiers, legal applicants and utility docket conditions were not established in this audit.

Risks and unknowns

Phased customer demand, tenant fitout, utility commissioning, construction labor and equipment delivery can alter the schedule. Contract quantities delivered, tariff obligations, signed leases, occupancy and project financing remain insufficiently disclosed.

What would change the assessment

A verified occupancy or commissioning record, delivered utility service, disclosed customer activation and project financing would strengthen readiness. An amended service arrangement, construction delay or evidence of reduced tenant demand would warrant reassessment.

Coverage and scope

Adds private wholesale operator diversity in Mississippi beyond AWS. One Compass Meridian campus; first two buildings under construction, later buildings remain planned.

Financial significance and assumptions

The announced $10 billion investment includes future tenants’ IT equipment and full campus fitout. It is not Compass-only construction spending. Project-specific financing commitments, loan closure, lease economics and expected returns were not established.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• SO: Mississippi Power utility parent; service economics are not campus compute revenue.

Government record audit

Mississippi Development Authority · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

not established

Record type

Tenant incentive announcement

Applicant / entity

Corderill LLC

Filed

not established

Decision

not separately established

Status

Incentive approval reported by MDA

Phase scope

Tenant equipment investment and employment incentive, not whole-campus permit

Conditions / limits

Employment expectation; full incentive instrument not retrieved. Campus permit identifiers not established.

Source basis and claim audit

6 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. Mississippi Power ↗ · Primary evidence. Published not displayed; event not separately established; reviewed October 7, 2026.

   Signed nearly 500MW long-term agreement; eight buildings/eight years; investment includes tenant IT; industrial-park substation and water/sewer/site grants described. Agreement terms, commissioning dates, full campus energization not disclosed.

2. Compass Datacenters / Mississippi Development Authority release ↗ · Primary evidence. Published 2025-01-09; event not separately established; reviewed October 7, 2026.

   Reproduces MDA statement, count as this underlying release once. MDA site assistance/operator certification plans; local shovel-ready infrastructure; $10B future occupied value. Contemporary ownership statement says Brookfield Infrastructure/Ontario Teachers/KKR backing, but obtain ownership-specific source if material.

3. Mississippi Development Authority ↗ · Government record. Published 2026-04-22; event not separately established; reviewed October 7, 2026.

   Corderill LLC announced as tenant, ≥$100M planned data equipment investment, ≥20 jobs. MDA approved Data Center Incentive conditional on expected employment. No GPU/hyperscaler ultimate customer identity or operational delivery established; do not guess Corderill's ultimate owner.

4. DatacenterDynamics ↗ · Independent reporting. Published 2025-02-04; event not separately established; reviewed October 7, 2026.

   Compass told DCD each planned building 250,000 sqft/40MW IT. Eight-building construction plan, $10B total includes tenant equipment. This is company-reported plan, not installed computing.

5. WTOK ↗ · Independent reporting. Published 2025-01-28; event not separately established; reviewed October 7, 2026.

   Onsite groundbreaking, site near interstate corridors next to Southern Pipe and Supply; single-tenant developer model explained by Compass VP.

6. WTOK ↗ · Independent reporting. Published 2026-07-21; event not separately established; reviewed October 7, 2026.

   Work begun first two buildings; first building anticipated end2026 (reported target ONLY); construction onsite, count attributed local release >1050 workers. No operation/occupancy certificate.

Permanent project profile →

Permanent project profile →

AVAIO Digital Taurus, Brandon, Mississippi

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

Taurus has moved beyond announcement through an executed electric-service agreement and company-reported sitework with a named contractor. Its changing occupancy target and enlarged capacity marketing warrant phase-specific treatment rather than treating the entire campus as ready.

Site and scope

AVAIO identifies a 329-acre site at East Metropolitan Center in Brandon, Rankin County. The initial announcement described 175 acres; DCD explicitly identified the discrepancy and sought clarification. The current 329-acre figure remains attributed to AVAIO.

Power basis

AVAIO says an Entergy Mississippi ESA was executed in August 2025. Its construction release specified 116 MW of Phase 1 power capacity, not independently verified IT capacity. The current page advertises a 1,036 MW long-term baseload envelope including 536 MW grid and 500 MW behind-the-meter supply; this is not verified energized or fully committed capacity.

Capital and financing

Promoters describe an initial $6 billion investment. No project-level financing close, lender terms, customer lease value or earnings contribution was established. The announced amount should remain an investment plan rather than money proven spent.

Demand and economic roles

AVAIO is the private developer, owner and operator; Yates Construction is the named general contractor; Entergy Mississippi is the utility. MDA provides incentive support and Rankin First supports the industrial site. Named customers, GPU vendors and contracted customer MW were not established.

Construction milestones

Strongest documented phase only; no whole-campus completion inferred. Company-reported sitework; September 24, 2026 local visit found clearing but no structures; August 2027 occupancy target

Government and utility record limits

The governor’s announcement records MDA tax-exemption approval. Rankin First’s detailed FAQ says the Entergy agreement was approved by MPSC and assigns interconnection costs to the customer. The MPSC order, docket identifier, approval date and full conditions were not retrieved; attribute that claim to Rankin First.

Water and community considerations

AVAIO describes water infrastructure, sewer extension work and recycled-water options. Rankin First distinguishes non-evaporative Phase 1 cooling from possible graywater use in future phases. WLBT’s original document-based reporting records Brandon residents’ clearing, grading and drainage concerns; those concerns do not establish a permit violation.

Risks and unknowns

Customer demand, financing, electric delivery, later behind-the-meter generation approvals and drainage conditions remain material uncertainties. Campus building and air permit identifiers were not established. Company-reported sitework supports construction-stage coverage, not a finding that every phase is fully permitted.

What would change the assessment

Retrieved permits and the MPSC order, measurable building progress, disclosed signed customer commitments and actual energization would strengthen the assessment. Further timetable changes, financing gaps or unresolved generation and water conditions would weaken it.

Coverage and scope

Adds distinct private Taurus development and utility-service execution in Mississippi. One Taurus campus at East Metropolitan Center; Phase 1 sitework does not establish approval of later generation.

Financial significance and assumptions

Promoters describe an initial $6 billion investment. No project-level financing close, lender terms, customer lease value or earnings contribution was established. The announced amount should remain an investment plan rather than money proven spent.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• ETR: Utility parent; executed service agreement reported by AVAIO, delivered capacity and earnings unverified.

Government record audit

Mississippi Public Service Commission · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

not established

Record type

Utility service agreement approval claimed by Rankin First

Applicant / entity

not established

Filed

not established

Decision

not established

Status

Company-reported sitework; September 24, 2026 local visit found clearing but no structures; August 2027 occupancy target

Phase scope

Entergy Phase 1 service agreement, not whole-campus generation approval

Conditions / limits

Order, legal applicant and conditions not verified.

Source basis and claim audit

7 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. AVAIO Digital ↗ · Primary evidence. Published 2025-09-10; event not separately established; reviewed October 7, 2026.

   Sitework underway, Entergy ESA executed August2025, Yates Construction GC; 329 acres, Phase1 >600k sqft/116MW power; historical H1 2027 delivery target.

2. AVAIO Digital ↗ · Primary evidence. Published undated live page; event not separately established; reviewed October 7, 2026.

   First occupancy now August2027,329acres; executed ESA for 2027; company says zoned; sewer extension underway and recycled water option; enlarged long-term power figures. No proof later generation/BTM permits/financing. Don't retain obsolete H1 target as current.

3. Office of Governor Tate Reeves ↗ · Government record. Published 2025-08-19; event not separately established; reviewed October 7, 2026.

   MDA Data Center Enterprises tax exemption approval/site grants and Entergy participation. Incentive is not construction completion. Confirm exact date header if displaying (search reports Aug19; initial OPEN excerpt omitted header).

4. City of Brandon ↗ · Government record. Published 2025-08-22; event not separately established; reviewed October 7, 2026.

   City confirms operator/site/AI-ready plan but historical175acres; promotions/quotes overlap initial release, don't count same shared statements as independent corroboration.

5. DatacenterDynamics ↗ · Independent reporting. Published 2025-09-13; event not separately established; reviewed October 7, 2026.

   Specialist reporting of sitework/Yates and independent editorial identification of175vs329-acre discrepancy; sought clarification. Underlying sitework claim company-reported, not reporter onsite.

6. Rankin First ↗ · Primary evidence. Published undated live FAQ; event not separately established; reviewed October 7, 2026.

   Distinct substantive local project disclosure: Phase1 draws116MW; says Entergy agreement MPSC approved/customer pays interconnection; building-permit process still applies; power, cooling and fiscal terms described. MPSC docket/date/order NOT independently retrieved. Attribute claimed approval to Rankin First; no docket invented. Closed-loop Phase1 and graywater future phases; marketing household comparisons/savings not independently established.

7. WLBT ↗ · Independent reporting. Published 2026-07-29; event not separately established; reviewed October 7, 2026.

   Obtained consultant documents via records request; Brandon residents raised clearing/grading/drainage concerns. Does not imply Jackson moratorium applies Brandon. Distinct original local scrutiny; only Brandon-specific facts useful here.

Permanent project profile →

Related Spotlight · what the dated site record establishes before 2027 →

Permanent project profile →

AWS Falls Township / Keystone Trade Center, Pennsylvania

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

AWS Falls Township combines industrial-site reuse, local approval and documented construction with phase-specific environmental approvals. It is distinct from the nuclear-adjacent Susquehanna campus. Later generator review and statewide permitting changes remain relevant delivery risks.

Location and scope

DEP identifies the campus at 600 Ben Fairless Drive, Falls Township, Bucks County, within the Keystone Trade Center redevelopment. Local planning described roughly 247 acres, ten structures and more than two million square feet. Later reports vary in geometry; avoid asserting an exact final building count without the adopted plan.

Power arrangements

AWS told local reporting the campus would use PECO service and fund required infrastructure upgrades. A PECO representative described a take-or-pay arrangement and June 2026 energization of the initial connection. This does not establish all-campus energization or delivered computing capacity; contracted campus MW was not verified.

Capital and financial significance

Amazon’s June 9, 2025 announcement concerns at least $20 billion across Pennsylvania campuses. It cannot be allocated wholly or equally to Falls Township. NorthPoint’s broader industrial redevelopment costs likewise are not AWS-campus capital spending. Project earnings and construction spending remain undisclosed.

Economic roles and customers

Amazon/AWS is the campus operator and computing provider; NorthPoint is the site developer; PECO provides grid service. Williams is associated with a gas-meter station for backup generators, not verified baseload generation. Morrisville Municipal Authority supplies water and wastewater infrastructure. Customer identities and equipment vendor contracts were not established.

Issued permits and documentary scope

DEP’s live dossier records July 2025 Air Quality Plan Approval 09-0261 to Amazon Data Services, Inc., a November modification and August 2025 NPDES permit PAD090088 A-3. These are agency-summary verified milestones; underlying PDF retrieval failed. Exact issuance days and full permit conditions were not audited.

Later application and regulatory risk

DEP records an April 2026 application, 09-0261A, for 280 natural-gas and three diesel backup engines. Its dossier documents a July 18–August 17 comment period. August 20 independent reporting said review continued after removal from fast-track permitting. A final approval was not established; generator capacity is not IT load.

Construction and delivery milestones

Construction is confirmed by township and independent reporting. Local July 2026 reporting attributed a Q4 2026 first-sections operating target to AWS. Initial grid-connection energization does not prove customer service; classify construction underway and treat commissioning targets as unverified.

Risks and unknowns

Additional approvals, environmental conditions, public opposition, grid economics, brownfield obligations and phased customer demand can affect delivery. Whole-campus power quantity, operational IT load, occupancy and financing detail remain unknown. Statewide rule changes do not by themselves prove this campus cancelled.

What would change the assessment

A final 09-0261A decision and conditions, occupancy and commissioning evidence, and actual customer service would strengthen readiness. Permit restrictions, utility amendments or a verified delay would prompt reassessment. Keep nuclear-power claims confined to the separate Susquehanna project.

Coverage and scope

Adds southeastern Pennsylvania industrial redevelopment beyond Susquehanna and Lancaster. 600 Ben Fairless Drive / Keystone Trade Center; separate from Susquehanna nuclear-adjacent campus.

Financial significance and assumptions

Amazon’s June 9, 2025 announcement concerns at least $20 billion across Pennsylvania campuses. It cannot be allocated wholly or equally to Falls Township. NorthPoint’s broader industrial redevelopment costs likewise are not AWS-campus capital spending. Project earnings and construction spending remain undisclosed.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• AMZN: AWS operator and compute provider; statewide $20 billion pledge is not campus-specific capex.
• EXC: PECO utility parent; initial connection is not delivered computing or quantified earnings.

Government record audit

Pennsylvania DEP · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

09-0261

Record type

Air quality plan approval

Applicant / entity

Amazon Data Services, Inc.

Filed

not established

Decision

July 2025; exact day not established

Status

Issued; modification November 2025

Phase scope

Earlier campus backup-generator approval

Conditions / limits

Agency summary verified; underlying permit PDFs inaccessible. Exact issuance day and full conditions not established.

Pennsylvania DEP · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

PAD090088 A-3

Record type

NPDES construction stormwater permit

Applicant / entity

Amazon Data Services, Inc.

Filed

not established

Decision

August 2025; exact day not established

Status

Issued per DEP summary

Phase scope

Construction stormwater

Conditions / limits

Agency summary verified; underlying permit PDFs inaccessible. Exact issuance day and full conditions not established.

Pennsylvania DEP · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

09-0261A

Record type

Air quality plan approval application

Applicant / entity

Amazon Data Services, Inc.

Filed

April 2026; exact day not established

Decision

not established

Status

Application; final decision not established

Phase scope

Proposed 280 gas and three diesel backup engines

Conditions / limits

Agency summary verified; underlying permit PDFs inaccessible. Exact issuance day and full conditions not established.

Source basis and claim audit

6 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. Amazon ↗ · Primary evidence. Published 2025-06-09; event not separately established; reviewed October 7, 2026.

   $20B statewide investment for Falls/Salem and additional locations; company/customer AWS; do not allocate capital equally or to one campus.

2. Pennsylvania Department of Environmental Protection ↗ · Government record. Published undated live dossier; event not separately established; reviewed October 7, 2026.

   Construction600 Ben Fairless; township March2025 approval; July2025 issued Air Plan Approval09-0261 to Amazon Data Services,Inc.; August2025 issued NPDES PAD090088 A-3; Nov2025 modification09-0261. April2026 application09-0261A280gas+3diesel backup units, public comment July18-Aug17. Later approval NOT established. Generator permit≠ITload or full campus completion. PDF links click unavailable; disclose summary-level agency verification only, exact issuance days and conditions not established.

3. Pennsylvania Department of Environmental Protection ↗ · Government record. Published 2026-07-13; event not separately established; reviewed October 7, 2026.

   Application09-0261A and backup-only use; confirms government hearing and proposed status. Distinct document but same agency not independent corroboration.

4. Falls Township ↗ · Government record. Published not established; event not separately established; reviewed October 7, 2026.

   Approved modified digital campus plan247acres10structures>2M sqft; NorthPoint developer; PECO design coordination attributed developer. Header date omitted excerpt; exact approval March2025 supported DEP, no exact day invented. $1.5B/$40-45M figures relate overall NorthPoint industrial redevelopment NOT AWS campus.

5. LevittownNow ↗ · Independent reporting. Published 2026-07-15; event not separately established; reviewed October 7, 2026.

   AWS says PECO sourced/upgrades funded AWS; PECO representative take-or-pay arrangement, June initial connection energization, studies; Williams gas meter station for backup engines; phased demand-dependent campus construction and Q4 2026 first sections target. Choose ~200word aggregate attributed content maximum. Power_connection energization not customer compute delivery. Some geometry/building figures differ other documents; don't assert exact final campus structures without adopted plan.

6. Philadelphia Inquirer ↗ · Independent reporting. Published 2026-08-20; event not separately established; reviewed October 7, 2026.

   (turn21view5), accessible full text. DEP spokesperson said latest generators still reviewed and1236comments received; Shapiro removed data centers from fast-track Aug18; phase approvals distinguished. Governor executive-order text not separately opened, so attribute regulatory effect/report; no inference campus prohibited/cancelled. No final generator decision retrieved.

Permanent project profile →

Permanent project profile →

Bell AI Fabric Hillside / TRU

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

Physical construction and a university anchor make this more concrete than a sovereign-AI pledge. Investment relevance comes through BCE's compute platform, with commercial utilization and power economics still unquantified.

Campus boundary

One distinct McGill Road campus. Excludes nearby Equinix KA1, operating Bell Mission Flats and an unfinalized second TRU proposal.

Power and readiness

Planned hydroelectric infrastructure; no verified energization or campus-specific binding power contract. Closed-loop cooling design does not mean zero energy demand.

Demand and customers

CFJC reports university statement that approximately half the computing capacity supports academic research and the remainder enterprise services; this is not verified contracted utilization.

Milestones

City permit issuance February 13, 2026; university construction announcement April 9; independent construction reporting June 4; late 2027 completion remains target.

Risks

Schedule, cooling electricity demand, commercial customer conversion, unquantified capital/returns, local opposition and uncompleted heat-reuse integration.

What would change the assessment

Occupancy authorization, utility energization, commissioned compute capacity, disclosed customer commitments and quantified project economics.

Unknowns

Project capex and financing terms; supplier contracts; power tariff/service commitment; delivered IT MW; occupancy and delivery date. Network-wide Groq relationship is not a campus-specific hardware contract.

Coverage and scope

Adds a distinct purpose-built sovereign AI campus in interior British Columbia. 1452 McGill Road Hillside campus; excludes operating Mission Flats, Equinix KA1 and unfinalized second TRU proposal.

Financial significance and assumptions

BCE/Bell is operator and AI platform provider; TRU Property Trust is landlord and academic partner; TrueNorth Sustainable Infrastructure Inc is lead developer per TRU. Castanet identifies Hillside Data Centre Inc as lessee/developer and iTel as partner. Corporate relationship between developer entities is not independently established. No contract value, standalone financing close, project earnings or return verified.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• BCE: Bell operator and AI platform provider; campus returns and capital costs unquantified.

Government record audit

City of Kamloops · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

DPM01081; BP054664

Record type

Property information report listing development application and commercial building permit

Applicant / entity

Hillside Data Centre; legal parcel PL EPP58422 LT 1 SEC 1 TWP 20 RGE 18 MER 6; PID 029-892-163

Filed

2025-10-22

Decision

2026-02-13

Status

B-COMMERCIAL permit issued; no occupancy status established

Phase scope

1452 McGill Road / Hillside Data Centre

Conditions / limits

Underlying permit conditions not retrieved; report is not a survey or completion certificate.

Source basis and claim audit

7 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. City of Kamloops ↗ · Government property/permit record. Published 2026-09-24; event 2026-02-13; reviewed October 7, 2026.

   Lists DPM01081 Hillside Data Centre and BP054664 B-COMMERCIAL application 2025-10-22 issued 2026-02-13. Does not establish occupancy or energization.

2. Thompson Rivers University Community Trust ↗ · Primary institutional disclosure. Published 2026-04-09; event 2026-04-09; reviewed October 7, 2026.

   Construction underway at 1452 McGill Road, final university design review completed, TrueNorth named lead developer, late2027 completion anticipated; future heat reuse capability.

3. Castanet / Josh Dawson ↗ · Independent original local reporting/interviews. Published 2026-05-12; event not separately established; reviewed October 7, 2026.

   TRU-owned property leased to Hillside Data Centre Inc, Bell operating partnership; university interviews discuss closed-loop cooling, energy use and incomplete future heat reuse; second TRU facility not yet formally discussed.

4. CFJC / Victor Kaisar ↗ · Independent original reporting/site photos/interviews. Published 2026-06-04; event 2026-06-04; reviewed October 7, 2026.

   Construction underway; university reports 30000-square-foot AI compute facility and approximate half academic capacity, remainder enterprise services.

5. Bell ↗ · Current company campus footprint. Published date not established; event not separately established; reviewed October 7, 2026.

   Hillside AI Centre at TRU explicitly 26MW planned and hydroelectric-powered; not delivered load or independent power verification.

6. Bell ↗ · Historical company disclosure. Published 2025-05-28; event 2025-05-28; reviewed October 7, 2026.

   Academic training/inference partnership; initial 2026 TRU timing superseded. Groq relationship is network-wide and not verified hardware allocation at this campus.

7. DatacenterDynamics / Dan Swinhoe ↗ · Specialist reporting largely following TRU disclosure. Published 2026-04-14; event not separately established; reviewed October 7, 2026.

   Summarizes construction/late2027 target, distinguishes nearby Equinix site. Not additional independent proof of construction.

Permanent project profile →

Permanent project profile →

Bell AI Fabric Sherwood

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

Verified ongoing construction, named compute tenants and a quantified Bell capex plan make this a consequential sovereign AI build. Cash returns depend on delivery, tenant ramp and grid economics; the uncommitted1.2GW vision should not drive readiness.

Campus boundary

One300MW RM Sherwood campus; potential900MW expansion retained only as conditional context and not a separate counted project.

Power and readiness

Original300MW project grid-supplied according to IAAC. Substation construction scope identified; no delivered customer service established. Proposed expansion uses partner natural-gas power under non-binding MOU.

Capital and financing

Company plans debt and cash financing for C$1.7bn initial construction. No standalone financing close established. Leverage-neutral claim is management expectation on run-rate adjusted EBITDA.

Milestones

DP26-005 municipal development agreement approval reported April20; work begun April21 per local reporting; IAAC June5 designation unavailable because activity substantially begun; structural shells October1; H12027 first-hall target.

Risks

Municipal drainage/acoustic/aquifer conditions, community opposition, commissioning and cooling/power delivery, capex and debt exposure, tenant economics. Conditional expansion requires customer/commercial commitments, permits and assessments.

What would change the assessment

Energization and first-hall commissioning, public service terms, actual tenant utilization and capex disclosure, verified compliance with development agreement conditions.

Unknowns

Operating IT load, occupancy authorizations, detailed power contract/tariff, tenant lease economics, phase completion, financing close and expansion permits. IAAC not-designated status is not environmental approval.

Coverage and scope

Adds Saskatchewan sovereign AI delivery and a new Canadian provincial location. One Sherwood campus, initial 300 MW scope; nonbinding generation expansion is not a second project.

Financial significance and assumptions

BCE/Bell direct campus/platform operator; CoreWeave and Cerebras tenants per Bell announcement (do not infer lease value/terms). Bird Construction TSX BDT lead construction partner, contract value undisclosed; SaskTel fibre and go-to-market partner. NVIDIA GPU exposure is through CoreWeave's stated compute, not an independently verified Bell procurement contract. George Gordon First Nation land/procurement/workforce relationship. C$50bn-plus refers potential expanded ecosystem including customer hardware and generation, not current Bell capex.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• BDT: Named lead construction partner; contract value, project margin and realized earnings undisclosed.
• BCE: Operator and planned C$1.7 billion construction investor; expansion headline includes other participants.
• CRWV: Named compute tenant; contracted pricing and campus revenue undisclosed.

Government record audit

RM of Sherwood · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

DP 26-005

Record type

Advanced development agreement notice

Applicant / entity

not established

Filed

2026-04-02

Status

Notice establishes permitted Light Industry use and planned April20 agreement review; actual council approval verified in independent reporting only

Phase scope

NW33 16 19 W2M, initial300MW campus

Conditions / limits

Notice references Zoning Bylaw18/17 Section8.1.2 and applicant executive summary; full adopted agreement not inspected.

Impact Assessment Agency of Canada · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

90514

Record type

Request for designation determination

Applicant / entity

Bell Canada

Filed

2026-05-08 and 2026-05-15 requests

Decision

2026-06-05

Status

Request for designation not applicable: activity substantially begun

Phase scope

Bell Canada300MW industrial AI campus,65-hectare agricultural site, grid supplied

Conditions / limits

Not an environmental approval or determination of no impacts. Linked response document failed retrieval; public registry summary accessible.

Source basis and claim audit

8 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. Bell ↗ · Original company disclosure. Published 2026-03-16; event 2026-03-16; reviewed October 7, 2026.

   300MW campus, Cerebras/CoreWeave tenants, SaskTel fibre/market partnership, C$1.7bn construction capex plan financed debt/cash, first hall H12027 target; no actual returns inferred.

2. RM of Sherwood ↗ · Municipal primary notice. Published 2026-04-02; event 2026-04-02; reviewed October 7, 2026.

   DP26-005 location/legal use and scheduled development agreement review, not itself adopted approval.

3. Impact Assessment Agency of Canada ↗ · Federal primary determination registry. Published 2026-06-05; event 2026-06-05; reviewed October 7, 2026.

   Physical activities substantially begun;300MW65ha grid-supplied project; designation cannot apply. Not environmental approval.

4. CKRM / Martin Oldhues ↗ · Independent original council reporting. Published 2026-04-20; event 2026-04-20; reviewed October 7, 2026.

   Council unanimously approved development agreement;16 conditions including drainage, acoustic testing and record drawings, permit suspension possible for unaddressed off-site impacts; protests and consultation concerns.

5. Bell and Bird Construction ↗ · Joint primary company disclosure. Published 2026-05-14; event 2026-05-14; reviewed October 7, 2026.

   Bird lead construction partner, ATAL architect, George Gordon site services, first phase H12027; Bird mirror same release not extra source.

6. Government of Saskatchewan ↗ · Primary government construction-site update. Published 2026-10-01; event 2026-10-01; reviewed October 7, 2026.

   Four first-phase shells taking shape, preparation for interior fitout,500-plus workers; Dynamo Electric substation scope; first hall H12027 target, not operational.

7. Canadian Press / Anja Karadeglija via MidlandToday ↗ · Independent original summit reporting. Published 2026-09-14; event 2026-09-14; reviewed October 7, 2026.

   Company construction progress statement and expansion announcement; government says300MW grid plus potential900MW BYOP. Not proof financing or expansion completion.

8. Bell ↗ · Primary expansion disclosure. Published 2026-09-14; event 2026-09-14; reviewed October 7, 2026.

   Explicitly NON-BINDING MOU for up to900MW additional natural-gas partner generation; phased development subject to customers, commercial agreements, permits, approvals and assessments.

Permanent project profile →

Permanent project profile →

Kao Data Harlow AI campus / KLON-03

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

A contracted AI expansion at an established UK campus, supported by construction evidence and a project-specific environmental decision, offers a more tangible delivery case than speculative UK greenfield announcements. Listed exposure differs sharply: Nebius pays for compute hosting; Infratil participates through ownership; Shell supplies electricity.

Power and readiness

The existing campus supply and renewed Shell electricity arrangement are documented. Downing plans a private-wire solar project supplying up to 40 MW; neither the solar announcement nor annual renewable certificates proves continuous carbon-free electricity or new phase grid headroom. The environmental permit covers emergency generators, not IT commissioning.

Capital and financing

Infratil reports construction progressing and materially all operating and under-construction capacity contracted at year-end. No KLON-03-specific debt amount, lease pricing or return is disclosed. Nebius’s £1.7 billion UK program is a national allocation, not Harlow capex. Kao-wide revenue and EBITDA must not be assigned to this campus.

Company economic roles

Kao Data develops and operates the campus. Harlow Operations Limited is the permit operator. Nebius (NBIS) is the contracted AI cloud customer; Infratil (IFT) is a listed investor; Shell (SHEL) is the electricity supply counterparty; Downing is the prospective solar developer. NVIDIA-ready design is not a verified GPU procurement contract.

Milestones

KLON-03 was announced March 12, 2025. Downing disclosed the solar arrangement November 18, 2025. Infratil’s May 26, 2026 results reported construction progressing and capacity from early 2027. The Environment Agency granted its variation June 3, 2026; the Nebius agreement followed June 8. Shell signed its renewed agreement July 1 (published July 2).

Risks

Commissioning and cooling delivery, tenant ramp and concentration, electricity tariffs and prospective solar permitting remain material. Historical campus design totals differ from the current operator profile; use phase-specific figures and retain that discrepancy.

What would change the assessment

Verified ready-for-service and energized customer delivery would strengthen the assessment. A revised customer agreement, delayed power or cooling commissioning, or new backup-operating restrictions would weaken it.

Unknowns

The exact Nebius allocation between KLON-02 and KLON-03, GPU delivery date, Harlow revenue and EBITDA, lease economics, actual utilization, solar PPA terms and commissioning, and phase-specific grid capacity are not established.

Coverage and scope

Begins U.K. coverage through a documented AI expansion at an established campus. One Harlow campus; KLON-03 construction readiness, earlier buildings operational, no duplicate phase entry.

Financial significance and assumptions

Infratil reports construction progressing and materially all operating and under-construction capacity contracted at year-end. No KLON-03-specific debt amount, lease pricing or return is disclosed. Nebius’s £1.7 billion UK program is a national allocation, not Harlow capex. Kao-wide revenue and EBITDA must not be assigned to this campus.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• NBIS: Contracted AI customer; campus hosting costs differ from Nebius revenue.
• IFT: Listed investor in Kao; owner-wide results cannot be assigned to KLON-03.
• SHEL: Electricity supply counterparty; contract value and campus earnings undisclosed.

Government record audit

Environment Agency · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

EPR/WE4588AB/V002

Record type

Bespoke environmental permit variation decision

Applicant / entity

Harlow Operations Limited

Filed

not established

Decision

2026-06-03

Status

Granted

Phase scope

Emergency standby generators KLON1/2/3

Conditions / limits

Routine monthly readiness tests on weekdays between 9 a.m. and 5 p.m.; air emission and performance monitoring and reporting. Read the decision with the full permit. Covers emergency combustion plant, not computing commissioning.

Source basis and claim audit

7 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. Kao Data ↗ · Company disclosure. Published 2025-03-12; event 2025-03-12; reviewed October 7, 2026.

   KLON-03 construction announcement: 17.6 MW planned AI capacity, hybrid direct-to-chip cooling and up to 130 kW racks are company design claims.

2. Infratil ↗ · Listed owner results. Published 2026-05-26; event 2026-05-26; reviewed October 7, 2026.

   FY26 presentation, PDF page 24 (printed 19): KLON-03 construction progressing; capacity expected from early 2027. Materially all operating and under-construction capacity was contracted at year-end. Revenue £48 million and EBITDAF £4.7 million are Kao-wide figures.

3. Environment Agency ↗ · Government decision. Published 2026-06-03; event 2026-06-03; reviewed October 7, 2026.

   Permit EPR/WE4588AB/V002 granted to Harlow Operations Limited June 3, 2026. Adds KLON-03 emergency generators and extra KLON-01/02 units. 152 MW combined thermal combustion input is not IT load.

4. Kao Data ↗ · Company commercial disclosure. Published 2026-06-08; event 2026-06-08; reviewed October 7, 2026.

   22 MW ten-year Nebius campus agreement; AI Cloud and Token Factory to be hosted at Harlow. The entire allocation is not attributed solely to KLON-03.

5. DatacenterDynamics ↗ · Independent specialist reporting. Published 2026-06-08; event 2026-06-08; reviewed October 7, 2026.

   Independent specialist reporting of the Nebius agreement and national investment plan. Historical building capacity figures differ from the current operator profile.

6. Downing Renewable Developments ↗ · Power counterparty disclosure. Published 2025-11-18; event 2025-11-18; reviewed October 7, 2026.

   Prospective private-wire solar development up to 40 MW with planned PPA; announcement does not establish solar commissioning or continuous firm electricity.

7. Kao Data / Shell agreement ↗ · Power supply disclosure. Published 2026-07-02; event 2026-07-01; reviewed October 7, 2026.

   Shell Energy UK signed renewed supply agreement July 1, published July 2. Kao-wide approximately 140 GWh annual supply since 2022 is renewable-certificate matched, not exclusive Harlow generation.

Permanent project profile →

Investment theme: Power and the AI Buildout →

Permanent project profile →

VIRTUS Marienpark / BERLIN1 campus

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

Berlin construction and a named lender facility make this more tangible than a sovereign-AI pledge. €210m development financing is verified as lender-reported; it is not the whole campus cost, revenue, or evidence that all planned capacity is leased.

Power and readiness

Incoming electrical supply and IT load are different measures. The DATA2HEAT relationship documents a planned heat reuse pathway. No signed project-specific firm electricity service agreement or customer energization record was verified.

Capital and financing

Standard Chartered reports a €210 million bespoke facility. VIRTUS’s July 2025 commentary described €900 million campus investment; that is an attributed capital plan, not the confirmed facility amount. Drawdowns, covenants and tenant leases remain undisclosed. September 2026 portfolio financing cannot be allocated entirely to Berlin.

Company economic roles

VIRTUS is the developer and operator and part of STT GDC. Standard Chartered (STAN) is an arranger and hedging bank. DATA2HEAT, a joint venture of Investa Real Estate and GASAG Solution Plus, plans heat reuse infrastructure. Lending exposure is different from campus ownership and operating rent.

Milestones

GASAG documented groundbreaking June 6, 2025; DCD reported construction June 10. VIRTUS’s July 4 commentary targeted Q3 2026; that elapsed target is not proof of completion. Standard Chartered disclosed financing February 5, 2026. DCD’s April 16 reporting expected Marienpark opening in 2027.

Risks

Grid and electrical commissioning, lease-up, construction costs, financing conditions and heat network delivery are unresolved. The lender’s 57.6 MW full-buildout figure conflicts with the current English operator webpage’s 62.6 MW; retain both source dates and scope rather than silently reconcile.

What would change the assessment

Named customer contracts and documented commissioning would strengthen the assessment. Power delays, funding conditions or unexplained capacity revisions would weaken it.

Unknowns

Project permit identifiers, firm grid service terms, financing close date and drawdowns, customer leases, rental rates and delivered AI capacity remain unverified.

Coverage and scope

Begins Germany coverage with Berlin construction and named project financing. One Marienpark campus, distinct from VIRTUS Wustermark; no completed-service finding.

Financial significance and assumptions

Standard Chartered reports a €210 million bespoke facility. VIRTUS’s July 2025 commentary described €900 million campus investment; that is an attributed capital plan, not the confirmed facility amount. Drawdowns, covenants and tenant leases remain undisclosed. September 2026 portfolio financing cannot be allocated entirely to Berlin.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Listed company connections

• STAN: Arranger and hedging bank for lender-reported €210 million facility; not operator ownership.

Government record audit

Project-specific permit identifiers and underlying conditions were not established in this audit. Construction and counterparty evidence do not establish all-phase authorization.

Source basis and claim audit

6 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. GASAG Group ↗ · Utility and heat counterparty disclosure. Published 2025-06-06; event 2025-06-06; reviewed October 7, 2026.

   June 2025 groundbreaking; four buildings and 57.6 MW planned IT capacity. DATA2HEAT is an Investa/GASAG joint venture planning waste heat reuse.

2. DatacenterDynamics ↗ · Independent specialist reporting. Published 2025-06-10; event 2025-06-10; reviewed October 7, 2026.

   Construction start; four 14.4 MW buildings planned; heat network and ownership context reported.

3. VIRTUS ↗ · Company construction commentary. Published 2025-07-04; event 2025-07-04; reviewed October 7, 2026.

   Company construction commentary describes €900 million planned investment and original Q3 2026 delivery target; elapsed target does not establish completion.

4. Standard Chartered ↗ · Lender disclosure. Published 2026-02-05; event 2026-02-05; reviewed October 7, 2026.

   €210 million bespoke financing facility. Standard Chartered acted as Mandated Lead Arranger and Hedging Bank. Lender describes 57.6 MW at full buildout.

5. DatacenterDynamics ↗ · Later independent reporting. Published 2026-04-16; event 2026-04-16; reviewed October 7, 2026.

   Later independent reporting expects Marienpark opening in 2027; the main article covers a separate Wustermark campus.

6. VIRTUS ↗ · Current company profile. Published date not established; event not separately established; reviewed October 7, 2026.

   Current English operator profile describes 62.6 MW across four buildings, conflicting with lender figure; discrepancy requires attribution.

Permanent project profile →

Permanent project profile →

Data4 Frankfurt-Hanau / Großauheim campus

Compare this project → · Track its catalysts →

Approximate locality marker. Construction-stage classification does not mean every planned phase is permitted, energized or serving customers.

Investment takeaway

A brownfield Frankfurt-market campus combines September 2025 construction evidence with an attributed capital plan above €2 billion and a project-specific first partial generator authorization. Its power reserves and city connection assumptions do not establish delivered compute capacity.

Power and readiness

Company documents describe 180 MW power reserves. Hanau’s climate model instead uses 180 MVA connection capacity with a prospective 230 MVA ceiling; these are not interchangeable with operating IT load. Stadtwerke Hanau and Hanau Netz provide relevant infrastructure, while the neighboring 30 MWe CHP feeds the local grid rather than providing an exclusive 180 MW campus supply.

Capital and financing

Data4 increased its announced campus investment plan from more than €1 billion to more than €2 billion by September 2025. Brookfield ownership is documented in specialist reporting. No Hanau-specific closed lender facility was verified; platform capital pledges must not be assigned to the campus.

Company economic roles

Data4 is developer and operator; Data 4 Germany S.à.r.l. is the legal generator applicant. Brookfield is an indirect owner. Stadtwerke Hanau and Hanau Netz have infrastructure roles. Mainova is a planned waste heat counterparty. Engie’s documented contract is for the neighboring CHP, not a verified campus EPC contract.

Milestones

Acquisition and preparation were disclosed in 2023. Mainova disclosed a signed heat integration letter of intent August 29, 2024. Groundbreaking occurred September 15, 2025. The government portal records a June 17, 2026 decision for first partial standby generator authorization. The historical 2025 first-module target does not establish completed service.

Risks

Phase-specific approvals, electrical delivery, tenant leasing, brownfield construction and commissioning remain material. The full generator grant attachment returned 404, limiting review of conditions, amendments and appeal provisions.

What would change the assessment

Readable permit conditions, verified substation energization, first service and customer contracts would strengthen the assessment. Grid delay, appeal or capacity reduction would weaken it.

Unknowns

Full environmental conditions, amendments and appeals, energized load, ready-for-service date, tenant leases, site EBITDA, closed financing and electricity contract terms remain unverified.

Coverage and scope

Adds Frankfurt-market brownfield construction with project-specific regulator evidence. One Hanau campus; first generator authorization covers M1-M4, not whole-campus customer service.

Financial significance and assumptions

Data4 increased its announced campus investment plan from more than €1 billion to more than €2 billion by September 2025. Brookfield ownership is documented in specialist reporting. No Hanau-specific closed lender facility was verified; platform capital pledges must not be assigned to the campus.

Our analysis: construction can precede customer billing and cash returns. No utilization, pricing, margin or project earnings model is assumed. Investment plans, financing facilities and money spent are different quantities.

Government record audit

Regierungspräsidium Darmstadt · Record / evidence ↗. Reviewed October 7, 2026.

Identifier

RPDA - Dez. IV/F 43.1-53 u 35.14/38-2023/1

Record type

First partial generator authorization under BImSchG

Applicant / entity

Data 4 Germany S.à.r.l.

Filed

not established

Decision

2026-06-17

Status

Authorization decision recorded by official portal; full grant attachment inaccessible (404)

Phase scope

Emergency generators for modules M1-M4 of a six-module campus

Conditions / limits

Application describes up to 360 operating hours per year and 457 MW thermal input. Do not present these as full granted conditions or IT load without reading the decision.

Source basis and claim audit

7 distinct reviewed documents. Primary evidence and independent reporting are identified below; syndicated releases count once. Sources establish only the stated claim, not whole-campus completion.

1. Data4 ↗ · Company construction disclosure. Published 2025-09-15; event 2025-09-15; reviewed October 7, 2026.

   September 2025 groundbreaking; 25-hectare site; company 180 MW power capacity; planned investment increased above €2 billion.

2. DatacenterDynamics ↗ · Independent specialist reporting. Published 2025-09-17; event 2025-09-17; reviewed October 7, 2026.

   Independent construction reporting; Brookfield-owned operator. Its 25-acre figure differs from primary 25 hectares and should not be copied.

3. Hessenschau/dpa ↗ · Independent public broadcaster. Published 2025-09-15; event 2025-09-15; reviewed October 7, 2026.

   Public broadcaster reports groundbreaking and roughly €2 billion announced investment; includes regional Frankfurt context.

4. RP Darmstadt / UVP portal ↗ · Government project procedure. Published date not established; event not separately established; reviewed October 7, 2026.

   Official portal records authorization decision June 17, 2026. First partial generator proceeding covers M1-M4: 64 diesel engines plus one house generator, 457 MW thermal input. Full decision attachment returned 404.

5. Mainova / Stadtwerke Hanau ↗ · Utility counterparty disclosure. Published 2024-08-29; event 2024-08-29; reviewed October 7, 2026.

   Signed letter of intent with Data4 for waste heat integration via heat pump. Neighboring 30 MWe CHP feeds the local grid; it is not exclusive campus power supply.

6. CityofHanau ↗ · Government climate planning study. Published date not established; event not separately established; reviewed October 7, 2026.

   2026 climate plan, printed page 99: Data4 180 MVA connection assumption with prospective 230 MVA ceiling. Model assumptions do not prove energization.

7. Data4 ↗ · Earlier company disclosure. Published 2023-02-15; event 2023-02-15; reviewed October 7, 2026.

   Acquisition and preparation of 25 hectares; historical 2025 first-module and 2032 full-campus targets cannot prove completion.

Permanent project profile →

Permanent project profile →

Stargate Abilene

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · Amounts in USD · Company claims attributed; analyst interpretations labeled.

The one-minute investment takeaway

Abilene is a documented AI workload story, not simply a land or construction announcement. Oracle provides the compute service used by OpenAI; Crusoe delivers the physical infrastructure. The investor test is whether additional capacity becomes billable compute quickly enough to justify its financing and equipment costs.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Oracle reports 75% of capacity delivered as of September 2026; remaining delivery pending Readiness source →

What the sources establish

• Operator / company disclosure Operating evidence. Crusoe’s September 30, 2025 disclosure confirmed the first phase running on Oracle Cloud Infrastructure, with early training and inference workloads and NVIDIA GB200 racks. Source ↗
• Operator / company disclosure Revised delivery target. Crusoe’s March 27, 2026 update moved the existing campus’s six-building second-phase target to end-2026. The earlier March 2025 disclosure had targeted mid-2026. The 1.2 GW figure describes planned total power capacity, not verified current IT load. Source ↗
• Operator / company disclosure Scope boundary. The adjacent 900 MW Microsoft development is a separate two-building campus, with its first building targeted for mid-2027. The combined 2.1 GW headline must not be assigned entirely to Oracle or OpenAI. Source ↗

Who receives the money?

Our analysis of the disclosed company roles. Revenue routes below are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Oracle reported FY2026 cloud infrastructure revenue of USD 18.1 billion and negative USD 23.7 billion free cash flow. These are company-wide results, not Abilene results. Our analysis: the relevant tension is rapid infrastructure revenue growth alongside upfront funding needs. Backlog is valuable only as delivery converts it into recognized revenue and cash; the campus’s share of either measure is unknown. Financial disclosure ↗

Milestones that would change the assessment

What could break the investment case?

Our analysis: The delivery target has already shifted. Slower power or equipment commissioning would defer billing while some costs continue. Concentrated compute demand, financing expense and faster hardware obsolescence could compress returns even if the campus is completed.

What would change our view?

Our analysis: We would weaken the economic thesis if commissioning slips without matching customer protections, if available compute grows faster than paid usage, or if realized pricing falls faster than unit costs. We would strengthen it with verified utilization, stable contracts and improving cash conversion.

What remains unknown Not established in this review

No verified current full-campus IT load, campus utilization, realized compute prices, specific lease rates or project-level profit. The full site’s planned power capacity, Stargate-wide spending commitments and Oracle’s company backlog are three different quantities.

This is project research, not a valuation model or a buy recommendation. Financial illustrations are assumptions, not management guidance. Review current company filings and valuation before making an investment decision.

Permanent project profile →

Buckeye Technology Park

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Buckeye is a land-and-infrastructure development opportunity at an early economic stage. The most valuable next evidence is a binding customer commitment paired with deliverable power. Acreage creates room to build, but it does not tell investors when revenue begins or which listed company benefits. The developer, future campus owner and computing customer may be different parties.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Developer lists owned land in development; operating capacity unestablished Readiness source →

What the sources establish

• Operator / company disclosure Current listing. Tract lists 2,019 acres, 1,800 MW of capacity and an owned-and-in-development status. It describes 22 parcels and by-right data-center use. These are developer specifications, not confirmation of energized IT capacity. Primary source →
• Operator / company disclosure Historical scope. The August 28, 2024 acquisition announcement described 2,069 acres, potential for up to 40 data centers and utility planning that could support 1.8 GW. The reviewed sources do not explain the acreage difference. Potential full build-out is distinct from financed construction. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: value is created by converting land into usable, contracted infrastructure. The central diligence bridge runs from zoning to utility delivery, then customer financing, construction and billing. The 1,800 MW headline cannot be multiplied by a generic construction cost and called committed spending. Investors also need to know which counterparty pays for substations, roads and water infrastructure, and whether the developer retains ownership or sells parcels.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Infrastructure delivery can lag customer requirements, and financing costs accrue before revenue. Tenant concentration can weaken bargaining power. Water, cooling and community constraints require project-specific evidence; neither adequate supply nor a shortage is inferred here.

What would change our view?

Our analysis: Our confidence would rise with executed tenant agreements, financed phases and dated utility commitments. It would fall if infrastructure spending proceeds without contractual demand or if delivery obligations require costs that expected proceeds cannot cover.

What remains unknown Not established in this review

Tenant identity, financing terms, infrastructure budget, water and cooling plans, current energized capacity, completion dates and public-company project earnings remain unestablished.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Microsoft Fairwater

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · Amounts in USD · Company claims attributed; analyst interpretations labeled.

The one-minute investment takeaway

Fairwater’s first facility has crossed from construction into operation. That changes the investment question: the key issue is now how effectively Microsoft monetizes an operating AI system, rather than whether a large building can be completed. The adjacent second facility remains a future development.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. First facility operational; second facility targeted for 2028 Readiness source →

What the sources establish

• Operator / company disclosure Operating milestone. Microsoft announced the first facility fully operational on June 23, 2026, after startup activities in April. Its second Mount Pleasant facility was under construction with completion scheduled for 2028. Source ↗
• Operator / company disclosure Investment boundary. Microsoft’s estimated USD 4.7 billion Wisconsin hyperscale construction spending spans 2024–2028. It is not a disclosed GPU budget, current-year expense or standalone cost of the first facility. Source ↗
• Operator / company disclosure Technical connection. The September 2025 design disclosure identifies a system interconnecting hundreds of thousands of NVIDIA GPUs for training and inference. Treat that as a disclosed architecture and scale description, not an independently measured October 2026 installed inventory. Source ↗

Who receives the money?

Our analysis of the disclosed company roles. Revenue routes below are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: Fairwater is strategically relevant to Microsoft’s AI capacity, but site-level revenue and earnings are not disclosed. Training may support future products rather than immediately billable third-party consumption. That creates a timing mismatch between upfront hardware and construction commitments and eventual revenue. A completed facility can improve capacity while initially lowering cash flow or increasing depreciation.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Large fixed commitments amplify the cost of underutilization. Hardware generations may change before assets earn back their costs. Operating constraints, customer concentration, power costs and local concerns can affect expansion or ongoing operation.

What would change our view?

Our analysis: The thesis weakens if new capacity does not relieve demand constraints or product revenue fails to cover increasing depreciation and operating costs. Verified paid utilization, efficient delivery and stable margins would provide stronger support than another capacity headline.

What remains unknown Not established in this review

No campus revenue, profit, paid utilization or verified current GPU inventory is established. The operator’s fully operational statement describes the first facility; it does not imply all future facilities are complete.

This is project research, not a valuation model or a buy recommendation. Financial illustrations are assumptions, not management guidance. Review current company filings and valuation before making an investment decision.

Permanent project profile →

Meta Hyperion

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · Amounts in USD · Company claims attributed; analyst interpretations labeled.

The one-minute investment takeaway

Hyperion combines two different investment models: Meta spends to improve its AI capabilities, while Entergy develops the power infrastructure needed to serve a large customer. The same growth story can create construction demand, a utility capital requirement and a technology-company cash outflow before it produces operating returns.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Under development; expanded campus and utility infrastructure planned Readiness source →

What the sources establish

• Operator / company disclosure Campus scope. Meta’s July 13, 2026 update expanded the plan to 5 GW of compute capacity and more than USD 50 billion of Richland Parish regional investment. It names DPR, Turner and Mortenson as general contractors. These are planned scale and regional investment figures, not current operating capacity. Source ↗
• Operator / company disclosure Earlier utility approval. Entergy’s August 2025 announcement described LPSC approval for three generation facilities: two targeted for late 2028 and one for end-2029. This approval should not be treated as blanket approval of every later expansion. Source ↗
• Operator / company disclosure Expanded power plan. Entergy’s March 2026 agreement outlined seven gas-fired plants totaling more than 5,200 MW, transmission and storage. The announcement describes a regulatory filing and Meta-funded infrastructure; it does not establish that all expanded assets are approved or in service. Source ↗

Who receives the money?

Our analysis of the disclosed company roles. Revenue routes below are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Entergy reported completing a USD 2.175 billion common-stock offering with a forward component in Q2 2026. That is company-wide financing, not a Hyperion-specific equity raise. Our analysis: utility expansion can increase the earnings base while requiring debt or equity funding. Investors should test returns per share after financing—not merely growth in capital spending. For Meta, the critical question is whether AI-enabled revenue or efficiencies exceed the cost of the build and operation. Financial disclosure ↗

Milestones that would change the assessment

What could break the investment case?

Our analysis: Mismatch between campus readiness and power availability could delay usage. Cost overruns, interest rates, financing dilution and changes in customer commitments can alter utility returns. At Meta, greater spending may precede demonstrable product revenue by years.

What would change our view?

Our analysis: The thesis weakens if service dates slip, customer protections do not cover stranded assets, funding costs outpace allowed returns, or Meta’s incremental AI benefit fails to justify the capital commitment. The stronger case requires both infrastructure delivery and measurable economic use.

What remains unknown Not established in this review

No full-campus operating date, installed GPU inventory, project revenue, current IT load or campus profit is established. Projected customer bill savings are not Entergy earnings. The 5 GW compute plan and 5,200 MW generation package are different measures.

This is project research, not a valuation model or a buy recommendation. Financial illustrations are assumptions, not management guidance. Review current company filings and valuation before making an investment decision.

Permanent project profile →

AWS Susquehanna campus

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · Amounts in USD · Company claims attributed; analyst interpretations labeled.

The one-minute investment takeaway

This is one of the clearest cases where an AI infrastructure story connects to a disclosed commercial contract. Talen sells nuclear energy under a long-term AWS agreement; Amazon uses power to support cloud infrastructure. The contract is broader than the adjacent campus, so its value cannot be assigned wholly to Salem Township.

Readiness basis: Strongest documented phase only; no whole-campus completion inferred. Talen reports revised AWS PPA transition in April 2026; regional supply ramp and campus load remain distinct Readiness source →

What the sources establish

• Disclosed supply agreement Contract scope. The June 2025 agreement targets 1,920 MW at full quantity through 2042, with extension options and full volume expected no later than 2032. It can serve the adjacent campus and other Pennsylvania locations. Talen’s presentation described approximately USD 18 billion of notional revenue over a 17-year contract. Source ↗
• Operator / company disclosure Updated transition schedule. Talen’s 2025 annual report, filed in 2026, described transition beginning after transmission work expected in spring 2026, with full front-of-the-meter transition expected in spring 2027. These are disclosed expectations; this review does not establish completion. Source ↗

Who receives the money?

Our analysis of the disclosed company roles. Revenue routes below are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

The USD 18 billion amount is a management contract estimate, not annual sales, present value or profit. Dividing it by 17 produces about USD 1.06 billion per year as arithmetic only; the actual ramp and escalators mean annual revenue need not be level. For context, Talen’s Q2 2026 release guided to USD 2.025–2.225 billion of company-wide adjusted EBITDA for 2026. Contract revenue and company EBITDA are not comparable profit measures, and this agreement’s isolated contribution is not supplied. Financial disclosure ↗

Milestones that would change the assessment

What could break the investment case?

Our analysis: The contract reduces some merchant-price exposure but introduces customer concentration and long-dated delivery obligations. Nuclear outages, transmission constraints, replacement-energy costs and financing can alter margins. Changes in wholesale prices also affect the value of committing generation long term.

What would change our view?

Our analysis: The investment case weakens if the ramp slows, delivery costs absorb the contracted premium, or plant availability deteriorates. It strengthens with verified volume growth, dependable generation and demonstrated free-cash-flow conversion.

What remains unknown Not established in this review

No campus-specific portion of the regional PPA, current campus computing load or standalone project profit is disclosed. Contracted MW cannot be translated directly into installed GPUs. Small modular reactor exploration is not an operating asset.

This is project research, not a valuation model or a buy recommendation. Financial illustrations are assumptions, not management guidance. Review current company filings and valuation before making an investment decision.

Permanent project profile →

Meta Sturgeon County: capital spending needs a productivity payoff

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: this is an owner-user computing investment, so a landlord rent model is the wrong test. Meta must convert large upfront spending into useful computing capacity that improves its products or reduces unit costs. Investors need to connect infrastructure delivery to monetization while keeping project costs in Canadian dollars.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Groundbreaking announced July 2026; planned capacity Readiness source →

What the sources establish

• Operator / company disclosure Announced scope. Meta’s July 8, 2026 release describes its first Canadian campus as a planned 1 GW AI-optimized facility and says completed investment will exceed CAD 13 billion. Neither is current operating capacity or spending already incurred. Primary source →
• Disclosed supply agreement Confirmed power agreement. Capital Power’s July 8, 2026 release discloses a greater-than-10-year agreement for 250 MW of capacity and energy, backed by its Alberta portfolio, with service anticipated in the back half of 2028. Capacity and energy payments are included, but prices are not disclosed. Primary source →
• Operator / company disclosure Cooling and local works. Meta describes planned closed-loop liquid cooling with dry heat rejection, and approximately CAD 60 million for local infrastructure improvements. These are company plans and commitments, not independently measured operating results. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: a campus can be technically successful and still earn an inadequate corporate return. Training benefits may take time to reach revenue, while inference can have recurring unit economics tied to usage and product adoption. Distinguish long-lived buildings from faster-obsolescing servers and accelerators. The investment headline cannot identify that split. Currency translation also changes reported US-dollar cost without changing local engineering scope. Capital Power has a separate supplier return: contractual payment stability must be weighed against generation costs and performance obligations. Its disclosed 250 MW agreement does not establish supply arrangements for the entire campus.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Buildings, generation and computing equipment may arrive on incompatible schedules. Hardware can age while waiting for usable power. Corporate AI growth does not prove this specific campus earns its cost of capital, and clean-energy matching does not alone establish hourly physical supply.

What would change our view?

Our analysis: Productive deployment with measurable unit-cost improvements would strengthen the case. Persistent idle capacity, rising enabling costs or replacement spending before expected benefits would weaken it. Strong supplier orders could coexist with weak owner returns.

What remains unknown Not established in this review

Phase schedule, present spending, current power delivery, IT versus total-site interpretation of the 1 GW figure, hardware vendors, contract pricing, remaining supply arrangements and campus-attributable benefits remain unestablished here.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

STAK Energy Campus: cheap gas needs an executable computing business

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: STAK proposes to bring computing to North Slope gas rather than move all that energy to a distant computing market. The opportunity is an integrated fuel, generation and computing business. Its challenge is proving that savings at the fuel and cooling layers exceed remote construction, pipeline, networking, staffing and financing costs. The project remains a proposal in the records reviewed, not an operating AI campus.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Proposed campus; preliminary land-lease decision reviewed; final authorization and construction unverified Readiness source →

What the sources establish

• Government preliminary record Location and lease stage. Alaska DNR’s June 17, 2026 notice describes a proposed 50-year lease on 715.4 acres about 26 road miles south of Deadhorse. It extended comments to July 17. A closed comment period does not establish a final lease decision. Primary source →
• Applicant plan in government record Applicant design and timetable. STAK’s March 12, 2026 development plan describes approximately 1 GW of onsite gas generation, new pipeline supply and modular HPC infrastructure. It targets late 2028 initial operation; its summer 2026 start intention is not verified construction progress. Generation output is not equivalent to IT load. Primary source →
• Government preliminary record Investment scope. DNR’s May 12 preliminary decision references approximately USD 500 million of construction capital. The record does not supply an independently validated, all-in funded budget reconciling generation, pipelines, computing hardware and every phase. Primary source →
• Applicant plan in government record Computing and geography. The applicant describes cloud computing and AI training among intended uses. The map pin uses the approximate center of the four pad corners in its plan, near 69.912° N, 148.859° W; it represents a proposed site, not an occupied building. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: underwrite the first deliverable phase before giving value to gigawatt-scale expansion. A gas resource in the region is not delivered turbine fuel: treatment, compression, rights of way and pipeline construction can become the critical path. A computing customer may tolerate distance for some batch workloads, but network diversity, latency and data movement still matter. Identify whether STAK is a landlord, power seller or computing-service operator; each has different capital intensity and exposure to hardware obsolescence. The USD 500 million reference cannot safely be divided by headline generation capacity to derive an all-in cost per IT MW. Nor should generation and customer IT capacity be added together. An attractive fuel price is only one input to a project return.

Milestones that would change the assessment

What could break the investment case?

Our analysis: This combines several development risks rather than merely adding servers to a powered building. Pipeline or turbine delays can strand computing hardware; hardware delays can leave generation capital underused. Cold ambient air may reduce cooling energy but does not remove permafrost, freeze protection, snow, remote maintenance or seasonal logistics costs. Air authorization, fuel availability and restoration security can change both timing and capital. A single anchor customer could make financing possible while concentrating credit and renewal risk. Claimed environmental benefits require measured fuel and emissions performance; gas-powered does not itself establish low-carbon computing.

What would change our view?

Our analysis: A financed first phase with binding fuel supply, enforceable customer payments, executable permits and independently reviewed capital estimates would strengthen the case. Unfunded generation ambitions, no customer offtake, unclear pipeline rights or a budget omitting major hardware and enabling works would weaken it. Milestone evidence matters more than increasing the headline size. Public-stock exposure should wait for documented counterparties and material contracts.

What remains unknown Not established in this review

No final lease decision, issued entry authorization, verified construction start, operating capacity, funded all-in phase budget, binding gas supplier, turbine award, computing tenant, GPU order, network-service contract or site revenue was established in this review. Late 2028 is an applicant target. We use approximately 1 GW from the linked applicant plan and do not adopt higher capacity figures from secondary summaries without reconciling scope and document version.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Google The Dalles: an established campus still needs productive computing

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: The Dalles adds a mature owner-user campus to the atlas. Google’s long local operating history reduces uncertainty about the site’s existence, but it does not establish today’s AI workload, current capacity or site return. The investor test is whether incremental computing improves product economics after energy, cooling and hardware replacement costs.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing Google campus; current capacity and workload mix undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Operating history. Google identifies The Dalles as the location where it began building and operating its own data centers in 2006. The location page also records a fourth-building construction milestone in 2015. Primary source →
• Operator / company disclosure Investment scope. The page says Google has invested more than USD 7 billion in Oregon since 2006. That statewide cumulative figure is not the budget for a new individual building, current annual spending or campus earnings. Primary source →
• Operator / company disclosure Resource diligence. Google links a site-specific water profile and water-stewardship material from its location page, but does not disclose current campus IT MW, utilization or an AI-specific share of computing. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: the corporate return comes through the services using the campus, rather than rent collected from an independent hyperscale tenant. Distinguish a facility expansion from a hardware refresh: both consume cash, but their useful lives and productive output differ. An established site can reuse common infrastructure, yet denser equipment may require power and cooling upgrades that eliminate some apparent savings. A long operating history is valuable evidence of execution, not a guarantee of an attractive return on the next dollar invested. Alphabet’s aggregate growth cannot be assigned to this campus without an attribution method.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Cooling and water requirements can change with density and operating conditions. An older building may need substantial modifications before accepting new hardware. Statewide investment and economic-impact estimates can make scale appear financially precise while leaving site returns unknown. Equipment may become obsolete faster than the buildings housing it.

What would change our view?

Our analysis: Disciplined phase spending, resource readiness and improved workload economics would strengthen the case. Rising retrofit costs, idle computing equipment or spending that outpaces monetizable usage would weaken it. We would upgrade AI attribution only with workload-specific evidence rather than assume all Google capacity is AI.

What remains unknown Not established in this review

Current IT and facility MW, occupancy-equivalent utilization, workload mix, new-phase budget, hardware orders, tariffs, measured efficiency and campus-attributable revenue or operating savings remain unestablished by the location page. Map coordinates represent the campus area, not a specific server hall.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Switch Citadel: value billable phases before multi-gigawatt ambition

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: the Citadel combines an established colocation campus with a much larger development envelope. The earnings case depends on what is delivered and billable now, followed by the return on each added phase. Full-buildout ambitions should be treated as options requiring capital, power and customers.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing campus; operator markets multi-gigawatt expansion potential Readiness source →

What the sources establish

• Operator / company disclosure Existing campus evidence. Switch’s opening announcement describes Tahoe Reno 1 as the first facility in the Citadel campus at the Tahoe Reno Industrial Center. It used design ceilings of up to 130 MW for that building and 650 MW for the campus. Primary source →
• Operator / company disclosure Newer scope. The current location page markets gigawatts of capacity upon completion and links a Q2 2025 expansion update. The newer ambition is not proof of current energized load and should not be added to the historical design ceilings. Primary source →
• Operator / company disclosure Connectivity role. Switch describes SUPERLOOP connections linking its regional facilities and major California markets. These are operator-described network capabilities, not disclosed connection volumes or tenant income. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: an operating campus can offer expansion efficiencies through established staff, network access and shared infrastructure. Those benefits do not make new power free or eliminate commissioning risk. Model common electrical and site spending consistently across phases. A higher-density offering also needs customer-specific evidence: marketing a technical ceiling does not establish that every hall can deliver it, or that all costs are included in ordinary rent. Recurring property and service contribution is distinct from customer computing revenue.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Changing design envelopes can create inflated totals if historical and current ambitions are combined. Electrical and cooling readiness may lag shell construction. Concentrated hyperscale demand can support growth but increase acceptance and renewal exposure. Network latency claims require workload-specific testing; they do not establish universal suitability.

What would change our view?

Our analysis: Verified new-phase rent commencement and adequate returns on shared capital would strengthen the case. Expansion commitments without customers, repeated electrical delays or expensive density upgrades without compensation would weaken it. The operator’s claimed industry superlatives are not used as investment evidence.

What remains unknown Not established in this review

Current commissioned and billable MW, tenants, lease pricing, new-phase investment, power contracts, measured facility efficiency, supplier awards and site earnings remain undisclosed in the cited pages. The map pin denotes the Tahoe Reno campus area rather than a specific phase.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Meta Eagle Mountain: separate the operating campus from the new expansion

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: Eagle Mountain supplies a recent expansion within an established campus. The existing operation and September 2026 growth announcement should be modeled separately. For Meta, the payoff is better product economics and computing efficiency; for utilities and energy counterparties, the relevant evidence is contract scope and payment terms.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing campus online since 2021; expansion announced September 2026 Readiness source →

What the sources establish

• Operator / company disclosure Recent expansion. Meta’s September 14, 2026 announcement says the campus is expanding and its investment in Utah will exceed USD 3 billion. It does not disclose the expansion’s standalone budget or IT MW. Primary source →
• Operator / company disclosure Existing operation. Meta’s Eagle Mountain fact sheet says the campus went online in 2021 and identifies work with Rocky Mountain Power on energy planning. Meta’s statements about paying energy costs are company assertions, not a site tariff analysis. Primary source →
• Operator / company disclosure Energy scope. The September release says Meta supports a 150 MW Faraday battery project owned and operated by Lydian Energy, alongside broader Utah renewable and storage additions. Those generation and storage figures are not campus IT capacity. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: an expansion can share site infrastructure and workforce while still demanding substantial new electrical and cooling capital. The cumulative Utah spending headline does not reveal the incremental return. Meta’s product benefits may be spread across its global system, making direct site earnings attribution difficult. For energy infrastructure, distinguish capacity in MW from stored energy in MWh and annual delivered energy. A battery can shift electricity delivery across hours, but its usefulness depends on duration, cycling and the contractual role it actually serves.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Existing operations can make new buildings appear more advanced than they are. Compute hardware, utility works and facilities may arrive at different times. Renewable matching does not itself establish hourly physical supply or low electricity costs. The campus may absorb capital without a separately measurable product payoff.

What would change our view?

Our analysis: Timely accepted capacity and improved computing efficiency would strengthen the assessment. Expensive upgrades, long idle periods or replacement spending ahead of benefits would weaken it. Named supplier or energy contracts could improve public-market attribution, but no investment conclusion follows from participation alone.

What remains unknown Not established in this review

Expansion IT MW, standalone budget, completion dates, actual load, equipment orders, utility pricing, battery payment terms and campus-attributable financial benefits remain undisclosed in the cited materials. Map coordinates denote the Eagle Mountain campus vicinity.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Meta Stanton Springs: operating resilience has an economic cost

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: Stanton Springs adds an operating owner-user campus whose resource relationships are documented more clearly than its site economics. The investment test is whether resilient water, power and computing infrastructure produces useful workloads at competitive cost. It should not be labeled a dedicated AI training cluster without specific evidence.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operating Meta campus; current IT capacity and AI workload share undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Documented operation. Meta’s April 3, 2026 article discusses operations at Stanton Springs in Newton County, Georgia. It identifies the Newton County Water and Sewerage Authority as the source of all campus water and says groundwater is not used in operations. Primary source →
• Operator / company disclosure Service relationship. The article quotes the water authority as confirming service agreements and operation within the agreed allocation. This is attributed evidence, not an independent measurement of daily withdrawal, tariff or drought resilience. Primary source →
• Operator / company disclosure Investment figures need reconciliation. Meta’s location pages and dated fact sheets publish differing investment headlines. We do not treat an unreconciled cumulative figure as the cost of a current expansion or an operating-income estimate. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: a water agreement can reduce operating uncertainty, but its cost and resilience need to be evaluated alongside electrical service and the intended computing equipment. Annual water restoration or community investment has a different economic role from delivering water at the hottest operating hour. Mature campuses also face continuing replacement and efficiency decisions. A new server generation may improve computing output while increasing local heat rejection and electrical requirements. The important denominator is useful work, rather than building area or a cumulative investment headline.

Milestones that would change the assessment

What could break the investment case?

Our analysis: An operating campus can still face resource, maintenance and hardware-obsolescence risk. Water-service compliance at one date does not establish indefinite capacity for future expansion. Cumulative investment figures can vary because of reporting dates and scope; picking the largest without reconciliation would exaggerate precision. Corporate AI growth cannot be assigned to the site without workload evidence.

What would change our view?

Our analysis: Measured efficient operation, clear expansion obligations and competitive workload costs would strengthen the case. Resource bottlenecks, heavy retrofit spending or unexplained capital growth without productive output would weaken it. Verified AI deployments would strengthen thematic relevance, while broader enterprise workloads could still support infrastructure value.

What remains unknown Not established in this review

Current IT MW, AI workload share, measured utilization, site-specific return, tariff terms, peak water demand, new-phase budget and named equipment contracts remain unestablished in the cited article. The map pin represents the Stanton Springs campus vicinity, with no claim to identify a particular building.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Equinix SV11

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

SV11 shows how connectivity and existing campus infrastructure can support customers without a new hyperscale development. The financial test is recurring deployments and service attachment relative to operating and upgrade costs. The original development budget and cabinet plan provide useful historical context, but do not establish current occupancy, AI demand or a new procurement cycle.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator-listed colocation facility; historical development plans distinguished from current results Readiness source →

What the sources establish

• Operator / company disclosure Current marketed footprint. Equinix lists SV11 at 5 Great Oaks Boulevard in San Jose with 121,930 square feet of colocation space, N+1 power and N+2 cooling redundancy. Floor area does not establish billable occupancy. Primary source →
• Operator / company disclosure Historical construction scope. The June 2, 2021 announcement described a USD 142 million facility, 1,450 initial cabinets and 2,950 at full build, with opening targeted for June 30. These are dated development figures, not current occupied-cabinet counts or an AI expansion budget. Primary source →
• Operator / company disclosure Power partner boundary. That release described expansion of Bloom Energy fuel cells toward 20 MW of campus capacity, intended as primary generation with utility and generator backup. It is a historical campus plan, not evidence that SV11 presently has 20 MW of operating IT load. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: the historical facility budget cannot be divided by current area and treated as today’s marginal cost. Later phases, power systems, customer fit-out and density changes may have different funding scopes. Connection revenue can supplement cabinet income, but network availability alone does not create a paid connection. For listed-company exposure, separate Equinix’s recurring infrastructure model from Bloom’s equipment and service economics.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Density upgrades may consume capital without enough additional billable demand. Power generation introduces fuel, maintenance and reliability considerations. Customers can choose cloud services or other facilities, limiting pricing and retention.

What would change our view?

Our analysis: Our confidence would rise with recurring deployments and economical upgrades, plus verified current power configuration. It would fall if capacity grows without service attachment or if generation and retrofit expenses absorb expected margins.

What remains unknown Not established in this review

Current occupied cabinets, IT MW, AI workload mix, current fuel-cell capacity, upgrade spending and site earnings remain undisclosed in the reviewed sources.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Empire AI Beta

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Empire AI is a research-computing procurement story with public-interest objectives. Equipment suppliers can receive commercial revenue even though the system is not a commercial cloud campus. Investors should distinguish a funded research phase, supplier revenue recognition and the consortium’s scientific outcomes. A queue of projects demonstrates interest, not billable cloud demand.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Beta reported fully online August 10, 2026; Gamma remains a future phase Readiness source →

What the sources establish

• Institution / consortium disclosure Operating evidence. The University at Buffalo’s August 10, 2026 report says Beta is fully online, with more than 300 research projects queued. It describes Gamma construction continuing toward completion by end-2027. Beta operation does not establish Gamma completion. Primary source →
• Institution / consortium disclosure Budget and equipment boundary. Empire AI describes Beta as a USD 40 million NVIDIA-powered supercomputer launched in 2026. That phase amount is not the consortium’s lifetime budget or a disclosed NVIDIA order value. Its comparison with Alpha describes capacity increases, not measured commercial revenue. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: hardware demand is real even when the end user does not sell compute commercially. However, project budgets can include networking, integration, facilities and services, so assigning every dollar to an accelerator vendor would overstate exposure. Renewed public or philanthropic funding can support future orders, but neither a queue nor successful research automatically guarantees a refresh budget. This project’s supplier effect must be evaluated relative to the supplier’s total business.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Operating costs and equipment obsolescence can exceed recurring funding. Access bottlenecks can leave research demand unmet despite installed capacity. Heat reuse depends on infrastructure delivery and usable thermal demand, and it does not establish a supplier profit margin.

What would change our view?

Our analysis: Our confidence in sustained procurement would rise with funded refresh plans and documented research utilization. It would fall if budgets are consumed by facilities or operations without additional computing procurement, or if delivery and access constrain useful work.

What remains unknown Not established in this review

Supplier-specific contract value, hardware allocation, operating budget, actual utilization, maintenance terms and Gamma procurement spending remain unestablished. Scientific benefits are not equated with shareholder profit.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Equinix MI6

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

MI6 offers existing infrastructure in the Miami market, where a connectivity narrative needs facility-level verification. A regional network gateway can support customer demand, but it does not establish that every site hosts the same services or AI workloads. The investment test combines billable deployments with the capital needed to preserve reliable operation and meet customer requirements.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing operator-listed facility; dedicated AI deployment unconfirmed Readiness source →

What the sources establish

• Operator / company disclosure Facility evidence. Equinix lists MI6 at 1525 NW 98th Court in Doral with 75,046 square feet of colocation space and N+1 UPS, generator and cooling redundancy. The published configuration does not quantify outage performance, leased capacity or current customer load. Primary source →
• Operator / company disclosure Demand boundary. The listing does not identify a dedicated AI training cluster, named AI tenant or new construction budget. Miami-wide connectivity claims require separate verification before assigning a specific cable connection or service deployment to MI6. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: reliability is an economic requirement because customer service obligations can expose the owner to costs or lost demand. Redundancy specifications alone do not establish those financial terms. Weather resilience, insurance and backup-service requirements are diligence questions rather than findings of a problem at MI6. Capital spent to sustain service should be separated from capital that adds new billable capacity.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Sustaining capital can rise faster than rent. Customer consolidation or competition can pressure retention. High-density requirements may necessitate upgrades without an adequate pricing premium. Regional network strengths do not automatically deliver site revenue.

What would change our view?

Our analysis: Our confidence would rise with durable customer expansion and stable cash after sustaining costs. It would fall if upgrade and service obligations consume most growth or if demand evidence remains solely regional.

What remains unknown Not established in this review

Current IT MW, utilization, customer mix, AI hardware, pricing, insurance and maintenance costs, upgrade budget and site earnings remain unestablished.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Equinix TR6

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

TR6 adds an existing Greater Toronto infrastructure case where customer expansion can matter more than new construction. Recurring relationships can support rent and connectivity revenue, but the investor test is how much cash remains after the spending needed to support them. The facility’s location and marketed footprint do not establish an AI-only deployment.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing operator-listed facility; current AI-specific economics undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Facility evidence. Equinix lists TR6 at 30 Bramtree Court in Brampton with 31,194 square feet of colocation space and N+1 cooling redundancy. The operator listing supports a physical facility offering, not current occupied cabinets or productive computing load. Primary source →
• Operator / company disclosure Revenue boundary. The reviewed page does not identify a dedicated training cluster, named AI customer, site revenue or new project investment budget. Toronto-wide infrastructure demand and Equinix partnerships cannot establish those missing TR6-specific facts. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: existing customers can add power, cabinets or connections, but expansion may require owner-funded improvements. The relevant comparison is net cash growth after those costs, not just a higher bill. Gross floor area cannot establish the sellable footprint or rent. Portfolio results and customer-retention measures offer context but cannot be assigned to this one facility without supporting disclosure.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Rising billings can reflect electricity recovery rather than improved profit. High-density requirements can require substantial retrofits. Customer migration or consolidation can reduce existing deployments even while regional AI demand grows.

What would change our view?

Our analysis: Our confidence would rise with durable customer expansion supported by economical upgrades. It would fall if capital and operating costs absorb the revenue increase or if demand evidence remains limited to general market claims.

What remains unknown Not established in this review

Current IT MW, billable cabinets, utilization, customer identities, AI mix, pricing, upgrade costs and site profit remain unestablished. No CAD project budget is invented.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Compass Hoffman Estates: redevelopment is only the first capital gate

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: the former Sears headquarters offers a large redevelopment opportunity, but demolition and a headline investment figure do not establish recurring earnings. The investment case depends on turning successive buildings into contracted, powered space while limiting capital committed ahead of customer demand.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Redevelopment announced; current energized and leased capacity undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Dated development evidence. Compass announced the redevelopment kickoff on June 13, 2024, describing five hyperscale buildings on nearly 200 acres. It acquired the property in September 2023. Primary source →
• Operator / company disclosure Investment scope. The announcement described approximately $10 billion in local investment once operational. That community investment headline is not a disclosed developer construction budget or earned revenue. Primary source →
• Operator / company disclosure Historical schedule. The release anticipated summer demolition and construction beginning the following year, and identified ComEd participation. Those dated intentions do not verify October 2026 completion or a specific power contract. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: this is a sequencing problem. Site preparation can consume cash before any rent begins, while a large campus allows later buildings to remain options rather than immediate spending obligations. Compare committed capital with the portion backed by enforceable leases, and separate developer infrastructure from customer computing equipment. A public-market proxy needs a documented ownership or supplier relationship plus attributable economics; proximity to the project is inadequate.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Upfront remediation, demolition and utility work may outrun leasing. A completed shell can still carry financing and tax costs without rent if electrical delivery or customer acceptance slips. Broad investment estimates can obscure which party pays for servers and which capital earns a property return.

What would change our view?

Our analysis: Signed tenant commitments, independently evidenced energization and phased rent commencement would strengthen the case. Repeated deferrals, speculative construction without leasing, or rising enabling costs would weaken it. Treat municipal enthusiasm as context rather than proof of stabilized economics.

What remains unknown Not established in this review

Current construction progress, energized MW, tenants, lease pricing, financing terms, allocation of the investment headline and project-specific supplier contracts remain unestablished by the cited announcement.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

DataBank SNA2: measure the return on incremental capacity

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: SNA2 offers a concrete completed-expansion case. The question is whether incremental space and power generated profitable leasing, and whether further workload upgrades require new capital. Regional technology employers are potential sources of demand, not confirmed tenants.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Expansion completion announced April 2023; current utilization undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Completed addition. DataBank’s April 12, 2023 release says the expansion added 24,000 square feet of raised floor and 4 MW, bringing SNA2 to 57,700 square feet and 8 MW of critical IT power. Primary source →
• Operator / company disclosure Property and market. The release locates SNA2 at 17400 Von Karman Avenue in Irvine and describes carrier-neutral interconnection. Its combined Irvine figures include SNA1 and must not be attributed entirely to SNA2. Primary source →
• Operator / company disclosure Technical design. DataBank describes 2N power and N+1 cooling. Its discussion of nearby technology firms does not identify those firms as customers or establish AI-specific utilization. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: existing sites may reuse staff, land and some common systems, improving expansion economics. That advantage must be tested against the actual additional electrical and cooling investment. Separate new-customer growth from customers relocating within the operator’s portfolio. Revenue migration can fill a new hall without generating equal net corporate growth. An older completed expansion also invites a renewal analysis rather than continuous reliance on its opening announcement.

Milestones that would change the assessment

What could break the investment case?

Our analysis: A completed hall can underperform if leasing takes longer than expected or service costs rise. Higher-density workloads may consume power faster than floor area, reducing usable cabinet inventory. Local employer growth should not be converted into imaginary tenant commitments.

What would change our view?

Our analysis: Stable billable utilization, renewals and low incremental maintenance spending would strengthen the case. Heavy retrofit capital, weak retention or concessions that offset lease-up would weaken it. Evidence of AI demand needs named workloads or contracts rather than a regional technology narrative.

What remains unknown Not established in this review

Current occupancy, AI workload share, rental rates, expansion budget, site margins, measured efficiency, tenant identities and incremental service receipts remain undisclosed in the completion release.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Vantage VA2: monetize a campus in phases, not all at once

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: VA2 should be evaluated as three potential delivery and revenue streams rather than a single 96 MW earning asset. The operator provides a clear design envelope, but its forward-looking campus wording does not establish which buildings are commissioned today. The decisive evidence is phase-level power, acceptance and rent.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator lists campus design scope; current phase delivery not established here Readiness source →

What the sources establish

• Operator / company disclosure Campus envelope. Vantage lists an 18-acre campus with three multi-story data centers, 800,000 square feet and 96 MW of critical IT capacity. Its overview uses future-tense language. Primary source →
• Operator / company disclosure Infrastructure design. The operator describes multiple electrical feeds, N+1 electrical redundancy, closed-loop chilled-water cooling and N+2 mechanical redundancy. Primary source →
• Operator / company disclosure Connectivity design. The campus offering includes carrier-neutral connectivity, six meet-me rooms and three fiber entry points. These are specifications, not evidence of network-service revenue or tenant occupancy. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: the building count matters because revenue commencement can be staggered while shared infrastructure is paid for early. Allocate common substation and site costs consistently across phases, and avoid calling early phases profitable by leaving shared costs elsewhere. A market with many nearby users can support leasing, but customer willingness to pay requires timely usable power and acceptable technical specifications. Address and market branding should not substitute for precise property and utility evidence.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Shared infrastructure can become stranded if subsequent buildings are delayed. Network diversity does not compensate for insufficient electrical capacity. Existing cooling specifications also need customer-level validation before assigning modern high-density AI economics. Financing costs can accumulate during the interval between construction spending and accepted delivery.

What would change our view?

Our analysis: Verified commissioning, predictable phase delivery and lease terms that compensate the owner for customer-specific capital would strengthen the case. Extended power delays, disputed acceptance or a mismatch between design and tenant demand would weaken it. A current phase schedule would materially improve confidence beyond the undated campus offering.

What remains unknown Not established in this review

Current delivery dates, energized and billable capacity, tenant identities, utility arrangements, capital commitments, lease rates, measured efficiency and site-level financing are not established by the operator overview.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Edged Aurora expansion

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Aurora combines a preleased expansion with a clear physical construction milestone. That reduces some uncertainty about demand and structural progress, while leaving commissioning, customer handover and capital recovery as the next investor tests. The first operating facility and second planned building must be modeled separately rather than treating the entire expansion as already producing rent.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Second building topped out June 2026; Q2 2027 operating target Readiness source →

What the sources establish

• Operator / company disclosure Delivery and demand. Edged’s June 4, 2026 announcement says its second Aurora building is structurally topped out, fully preleased and planned for 72 MW of critical capacity. It targets operation in Q2 2027 and names FCL Builders. The tenant is not identified. Primary source →
• Operator / company disclosure Cooling and phase boundary. The release names ThermalWorks cooling and says the first Aurora facility opened in February 2025. Its water-savings comparison is an operator estimate versus conventional facilities, not measured second-building operation. Waterless cooling is not a claim of zero total site water use. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: preleasing improves demand visibility but does not remove execution or counterparty risk. Structural completion precedes electrical, cooling and computing acceptance. The efficiency case must also establish who pays for energy and who retains savings under the lease. Portfolio-wide design metrics are not a measured Aurora result. A planned 72 MW facility is therefore a delivery obligation before it becomes a proven cash-generating asset.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Power or mechanical readiness can lag the structure. Cooling must meet contracted density and reliability. Unnamed tenant concentration and financing costs can affect owner cash even when the building is preleased.

What would change our view?

Our analysis: Our confidence would rise with commissioned capacity and timely rent commencement at expected costs. It would fall if fit-out or utility requirements delay handover, or if concessions and upgrades absorb the planned efficiency benefit.

What remains unknown Not established in this review

Tenant identity, lease terms, project budget, financing, current commissioned second-building MW, measured efficiency and partner contract values remain undisclosed.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Vantage CA3: delivered capacity still needs an earnings bridge

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: CA3 has stronger delivery evidence than a proposed campus. Its investment question has shifted from whether a building can open to how much of its capacity earns rent, at what incremental cost, and with which customer density requirements. Opening is a useful milestone, not a utilization disclosure.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Opened to customers October 2024; current utilization undisclosed Readiness source →

What the sources establish

• Contractor disclosure Delivered facility. DPR describes a four-story, 486,000-square-foot single-building campus on nearly eight acres and says it opened to customers in October 2024. Primary source →
• Contractor disclosure Capacity and utility. DPR reports more than 64 MW of critical IT capacity and identifies Silicon Valley Power as the electricity provider. Capacity is not a measurement of customer electricity consumption. Primary source →
• Contractor disclosure Engineering scope. The contractor describes an onsite substation, multiple utility feeds, N+1 electrical redundancy and closed-loop cooling with N+2 mechanical redundancy. These specifications do not disclose GPU deployment or rack-level liquid cooling readiness. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: delivered infrastructure reduces one category of risk but can leave leasing, concentration and retrofit risk. Model contracted capacity separately from billable capacity and actual electrical draw. A take-or-pay reservation may earn rent before full customer load, whereas unfinished tenant work can postpone billing. Those possibilities require lease evidence. A dense urban building also needs scrutiny of vertical circulation, structural loads and cooling distribution when customers change hardware generations.

Milestones that would change the assessment

What could break the investment case?

Our analysis: A high-capacity building can remain economically underfilled. Tenant-specific upgrades may erode an attractive headline rental rate, and a few large customers can concentrate renewal exposure. Renewable electricity claims do not by themselves establish physical hourly supply, tariff stability or low cooling costs.

What would change our view?

Our analysis: Sustained billable occupancy and tenant acceptance without major incremental capital would strengthen the assessment. Expensive retrofits, concessions, customer concentration or long gaps between reservation and billing would weaken it. Contractor delivery evidence should not be stretched into owner profitability.

What remains unknown Not established in this review

Current customer utilization, contracted and billable MW, pricing, site capital cost, lease structure, measured PUE, actual electricity draw and workload-specific cooling approvals remain undisclosed in the cited project account.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Compass Loudoun County: a capacity listing needs a cash-flow bridge

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: the Compass market listing establishes a campus offering, but leaves the key earnings bridge incomplete. Investors need to know how much capacity is usable, contracted and billable, and which capital obligations remain. A large number on a market page cannot supply those answers.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator lists a 79 MW campus; current energized utilization undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Operator-listed scale. Compass’s market page identifies a 79 MW hyperscale campus in Loudoun County, Northern Virginia. It does not label that figure as measured current customer draw. Primary source →
• Operator / company disclosure Evidence boundaries. The Northern Virginia entry describes development with local officials but does not name campus customers, provide building-level delivery dates or disclose project contract values. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: hyperscale facilities can have concentrated customers and custom delivery obligations. A long contract may provide predictable receipts but also impose significant upfront fit-out and acceptance risk. Separate the asset’s development yield from the equity return after financing. A leased phase can look attractive before interest and replacement reserves yet provide much less distributable cash. Public investment exposure requires a documented relationship and meaningful ownership economics.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Customer concentration can magnify renewal and credit exposure. A power-ready facility may still await customer acceptance, and pass-through electricity receipts can inflate revenue without equal margin. Common market branding can also conceal different phases with different delivery risks.

What would change our view?

Our analysis: Verified rent commencement and stable customer obligations would strengthen the case. A large gap between listed capacity and usable contracted capacity, or owner-funded spending without compensated rent, would weaken it. Neither interpretation can be resolved from the market listing alone.

What remains unknown Not established in this review

Tenant identities, lease terms, energized and billable MW, delivery schedule, project capital, utility agreements, financing and site earnings remain unknown in the cited entry.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Digital Realty ACC6

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

ACC6 is an existing facility in a major infrastructure market, so the investment question is how well the owner converts established space into recurring cash. The operator and building specifications are documented. A regional AI demand narrative does not reveal the facility’s customer mix, rents or upgrade needs, and therefore cannot substitute for lease evidence.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator-listed facility; site-specific AI demand and utilization undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Facility scope. Digital Realty lists ACC6 at 44461 Chilum Place in Ashburn, with a one-story, 262,000-square-foot building and N+2 UPS and cooling redundancy. These are operator specifications; they do not identify usable IT MW or billable occupancy. Primary source →
• Operator / company disclosure Demand boundary. The reviewed listing does not disclose a named AI tenant, facility revenue, current utilization or a project-specific accelerator procurement. Its Northern Virginia innovation-lab promotion does not establish an ACC6 customer deployment. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: an existing property can benefit from renewals and customer expansion, but renewal pricing must be evaluated after concessions, fit-out spending and service costs. Redundancy supports reliability but does not prove modern high-density suitability without additional engineering evidence. Portfolio financial results can provide context; ACC6’s individual contribution remains unknown.

Milestones that would change the assessment

What could break the investment case?

Our analysis: High-density changes can require expensive retrofits or reduce usable capacity. Lease renewals may involve concessions. Existing infrastructure can retain tenants yet still consume more sustaining capital than the rent growth supports.

What would change our view?

Our analysis: Our confidence would rise with customer expansion that produces stronger cash after upgrades. It would fall if incremental rent requires disproportionate investment or if AI demand remains a regional claim without facility-level evidence.

What remains unknown Not established in this review

Current IT MW, occupancy, customer identities, AI workload mix, rent and renewal terms, upgrade budget and facility earnings remain unestablished. Stock valuation requires company-level work beyond this project profile.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

CyrusOne Aurora expansion: track the new campus separately

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: the new Aurora development is a separate underwriting case from CyrusOne’s nearby operating campus. Existing local presence may help execution, but it does not make the expansion operational or preleased. Follow the new buildings’ capital, power and acceptance milestones independently.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. January 2025 groundbreaking; subsequent commissioning not verified here Readiness source →

What the sources establish

• Operator / company disclosure Correct initial scope. CyrusOne’s January 28, 2025 announcement identifies two buildings totaling 446,000 square feet and an initial 40 MW of IT capacity, with scalability for future growth. This corrects the prior 50 MW note. Primary source →
• Operator / company disclosure Separate campus. The announcement calls this its second Aurora campus and distinguishes a nearby existing three-building campus operating since 2016. Primary source →
• Operator / company disclosure Utility participation. ComEd participated in the groundbreaking and discussed providing electrical infrastructure. The release does not specify an energized date, tariff or signed utility service capacity. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: an initial capacity figure is a construction and leasing envelope, not a revenue forecast. Expansion potential should receive value only after incremental power, permits, spending and demand are established. An existing neighboring business may provide operational experience, but shared staff or infrastructure savings require evidence. Avoid attributing statewide corporate investment totals to this one expansion or counting nearby operating capacity twice.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Confusing the operating campus with the new development can conceal unfinished construction and exaggerate current earnings. Power infrastructure delivery, customer fit-out and financing may move on different schedules. Scalability can require substantial new capital rather than being a free addition to the initial facility.

What would change our view?

Our analysis: New-campus energization, tenant acceptance and durable lease economics would strengthen the case. Unclear property identity, repeated delivery delays or expansion spending without committed customers would weaken it. The most valuable update is a dated operational milestone with its specific capacity and campus identity.

What remains unknown Not established in this review

Post-groundbreaking progress, current energized MW, tenants, preleasing, project budget, completion dates, rental terms and binding utility arrangements remain unverified by the cited release.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Digital Realty EWR12: redundancy is a product, not an earnings forecast

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: EWR12 connects investors to a named public operator, but the facility listing supplies engineering specifications rather than an earnings estimate. The valuable question is how reliability supports pricing and retention after accounting for the capital and operating costs of redundant infrastructure.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator-listed facility; site-level utilization and operating MW undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Property identity. Digital Realty lists EWR12 at 365 South Randolphville Road in Piscataway Township, with a two-story, 323,000-square-foot building. Total building area is not rentable IT area. Primary source →
• Operator / company disclosure Redundancy offering. The facility page specifies 2N UPS redundancy and 2N cooling redundancy. It does not disclose present IT MW, occupied capacity or a named AI customer. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: two complete redundant paths can improve service continuity but also tie up capital and require maintenance. The owner must recover those costs through pricing, retention and efficient operations. Building scale alone cannot distinguish occupied revenue-producing rooms from support areas. Even a profitable facility may be small relative to a diversified listed company, so site connection and stock valuation should be assessed separately.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Legacy infrastructure may require modernization to meet changing customer density needs. A technically robust building can still face leasing concessions or expensive replacements. Renewable-energy percentages do not establish actual hourly supply, customer consumption or the site’s electricity cost.

What would change our view?

Our analysis: Durable renewals and upgrades recovered through contracts would strengthen the assessment. Rising maintenance burdens, unused powered rooms or pricing concessions would weaken it. A public operator relationship is useful for research but does not establish a favorable stock entry price.

What remains unknown Not established in this review

Current occupied and billable capacity, rentable IT space, operating MW, site income, maintenance capital, customer identity, workload mix and project-specific supplier contracts remain undisclosed in the facility listing.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Compass Etobicoke campus

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Etobicoke illustrates why an old construction target is an evidence lead rather than a current operating conclusion. Compass documented a land purchase and phased plan in 2020. The next underwriting step is to establish what was delivered, what is leased and which obligations remain. A company’s wider institutional backing does not establish this project’s financing or cash return.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Historical land purchase and phase plan; current delivery unestablished by reviewed source Readiness source →

What the sources establish

• Operator / company disclosure Historical development. Compass’s May 5, 2020 release confirms land purchase in Etobicoke and describes a first building planned for 30 MW of critical IT load, with 10 MW targeted for Q3 2020. The elapsed target date is not proof that delivery occurred. Primary source →
• Operator / company disclosure Budget and company boundary. The release quotes a phase-one dollar amount without an explicit currency denomination; we do not classify it as CAD or USD. The current company description on the page names institutional backers, but does not establish their site-level equity, debt or tenant contracts. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: a land acquisition, construction plan and commissioned lease are different economic stages. Six years after a dated target, the relevant action is to find fresh operating evidence rather than assume completion or infer a delay. Unclear currency also prevents a credible cost-per-MW calculation. Historical plan values are useful for tracing scope, while current returns require actual cost, funding and contractual cash.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Incomplete current evidence makes revenue timing uncertain. Utility delivery or customer requirements can change between planning and handover. Corporate capital availability does not ensure a particular phase earns attractive returns.

What would change our view?

Our analysis: Our confidence would rise with fresh commissioning evidence and contracted cash flows. It would fall if substantial capital obligations remain without verified demand or if essential scope and delivery information remains unresolved.

What remains unknown Not established in this review

Current operating status and IT MW, leased capacity, tenant identities, explicit budget currency, final cost, financing, supplier contracts and campus earnings remain unestablished by the reviewed release.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Digital Realty YYZ10

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

YYZ10 is an existing infrastructure offering whose investor relevance depends on what customers actually lease and pay for. Its Toronto-area location and operator connection are documented. The next step is to establish contracted demand and the cost of meeting it. Square footage and redundancy specifications provide physical context; they cannot substitute for an earnings model.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator-listed facility; AI-specific demand and site economics unestablished Readiness source →

What the sources establish

• Operator / company disclosure Physical scope. Digital Realty lists YYZ10 at 371 Gough Road in Markham with a one-story, 104,500-square-foot building, 2N UPS redundancy and N+1 cooling. The page describes deployments from single cabinets to multi-megawatt suites. Primary source →
• Operator / company disclosure Evidence boundary. The reviewed facility page does not identify a dedicated AI cluster, campus investment amount, current operating IT MW, utilization or customer-specific contract. Its reliability specifications establish a marketed infrastructure configuration, not productive computing demand. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: the distinction between contracted backlog and commenced rent is central. Space can be available, contracted, fitted out and billing at different times. Customer-funded fit-out and owner-funded upgrades can produce very different cash requirements. At the company level, evaluate recurring cash generation after maintenance and funding, including dilution from new equity. This single facility is not a proxy for Digital Realty’s entire portfolio or its AI growth.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Density retrofits can cost more than expected and reduce usable space. Customer start dates or renewals can disappoint. Power pass-through may lift reported revenue without improving margin. Higher financing costs can offset stronger leasing, even if a facility performs operationally.

What would change our view?

Our analysis: Our confidence would rise with commenced leases and customer expansion supported by economical upgrades. It would fall if substantial new capital is needed without matching contracted returns or if demand remains unverified despite broad AI marketing.

What remains unknown Not established in this review

IT MW, occupancy, customer identities, AI-specific usage, lease terms, facility earnings and upgrade budget remain unestablished. No equipment supplier or attractive stock valuation is inferred from the operator connection.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

QTS New Albany 1: establish delivery before estimating returns

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: New Albany 1 requires a building-level evidence chain. The operator provides property identifiers and a campus offering, but not enough information to calculate present rent or utilization. The next useful disclosure is a dated delivery milestone tied to a specific building and power quantity.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Operator lists DC1 and DC2; commissioned capacity not established here Readiness source →

What the sources establish

• Operator / company disclosure Buildings identified. QTS’s New Albany 1 listing identifies DC1 at 1235 Beech Road SW and DC2 at 1225 Beech Road SW. The page uses differing locality labels in repeated address blocks, so the street addresses are the clearer identifiers. Primary source →
• Operator / company disclosure Financial scope discipline. The page describes a USD 3 million commitment over three years to an Ohio household electricity-assistance program. That state-level commitment is not this campus’s construction budget or utility connection cost. Primary source →
• Operator / company disclosure Delivery gap. The listing does not provide commissioned MW or a dated building-level completion announcement, so operating capacity remains unestablished here. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: two buildings may share enabling costs but enter service at different dates. Model each lease commencement separately while including a fair allocation of roads, substations and other common works. A community contribution has a different purpose and scale from construction spending. Investors should avoid both counting that contribution as capital evidence and treating public relations activity as evidence of tenant demand.

Milestones that would change the assessment

What could break the investment case?

Our analysis: A campus listing can be mistaken for proof that both buildings are fully commissioned. Utility and customer schedules may diverge, leaving completed capital idle. Regional infrastructure enthusiasm does not establish this property’s customer concentration, tariff or return.

What would change our view?

Our analysis: Dated building completion and enforceable customer commitments would strengthen the case. Unresolved utility timing, speculative fit-out or escalating shared costs would weaken it. Current operating evidence would improve confidence more than a broader market growth statistic.

What remains unknown Not established in this review

Present building status, energized and billable MW, tenants, budget, lease rates, financing, utility capacity agreements and construction awards remain unestablished by the cited listing.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

DataBank LGA3: nearly leased is stronger than merely built

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: LGA3 has both delivery and dated leasing evidence, making the financial question more specific than for a proposed campus. Investigate lease commencement, margins and concentration. The nearby LGA4 growth story remains a separate investment decision and should not be included in LGA3’s existing earnings.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Fully operational and nearly fully leased at May 2025 opening; current metrics undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Dated operating evidence. DataBank’s May 22, 2025 opening announcement calls LGA3 fully operational and nearly fully leased, including a leading cloud platform that it does not name. That statement is not an October 2026 utilization reading. Primary source →
• Operator / company disclosure AI connection. The announcement connects the facility to high-performance and accelerated computing and AI demand, but supplies no customer equipment inventory or site income. Primary source →
• Operator / company disclosure Project boundaries. DataBank identifies AECOM/Tishman as general contractor, locates the facility at 2000 Corporate Drive and discusses future LGA4 development separately. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: leasing reduces demand uncertainty, but “nearly fully leased” can include capacity whose rent starts later or requires additional owner-funded work. One large customer may create efficient operations and predictable cash flows while increasing credit and renewal concentration. Separate initial development profit, recurring property cash generation and any future expansion spending.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Heavy dependence on a cloud tenant can expose the owner to concentrated renewal negotiations. High-density service can demand continuing capital even after opening. Counting LGA4 as delivered capacity would exaggerate current scale and conceal future funding requirements.

What would change our view?

Our analysis: Stable billable utilization, retention and well-compensated technical upgrades would strengthen the case. Deferred rent, expensive owner-funded changes or concentrated customer departures would weaken it. An opening ceremony is historical delivery evidence, not a recurring financial report.

What remains unknown Not established in this review

Current occupancy, billable MW, rental rates, tenant identity, site capital cost, operating margin, contractor award economics and LGA4 commitments remain undisclosed in the announcement.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

QTS RIC5 expansion: permitting is the next investment gate

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: RIC5’s proposed additions should be evaluated as a permitting and development option. Existing buildings in the broader Richmond campus do not prove these additions are operating. The sponsor’s AI workload description supports relevance, while environmental authorization, usable power and capital commitments determine whether the proposal becomes an earning asset.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Proposed DC4–6 expansion in permitting; sponsor targets January 2028 construction start Readiness source →

What the sources establish

• Sponsor submission in government record Proposed additions. The sponsor-submitted FAST-41 notice proposes three additional buildings, DC4–6, south of Route 60 with associated utilities and supporting works. It describes the wider campus as serving a single AI-workload tenant without naming that tenant. Primary source →
• Sponsor submission in government record Permit boundaries. The notice references VDEQ approval dated October 10, 2025 for earlier RIC5 impacts and seeks expanded limits. Existing state authorization should not be treated as final authorization for the new proposal. Primary source →
• Sponsor submission in government record Schedule and investment threshold. The sponsor targets January 2028 construction and states investment will exceed USD 200 million, without a detailed budget. Broader-campus operations and ratepayer-protection statements do not independently verify the additions or utility cost treatment. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: early permitting work preserves a development opportunity but does not justify booking future income. Environmental mitigation can change usable land, layout, cost and delivery timing. A single-tenant development may reduce leasing uncertainty only if enforceable obligations cover the proposed buildings. Keep capital for existing campus operations separate from incremental expansion capital.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Permitting conditions can change scope and cost even when a project remains viable. Broad campus operations can obscure the early stage of the additions. Sponsor claims about ratepayer protection require actual agreements and regulatory evidence, not merely repetition.

What would change our view?

Our analysis: Final authorizations, transparent mitigation costs and tenant-backed funding would strengthen the case. Scope reductions, unresolved environmental impacts or power timing beyond construction would weaken it. FAST-41 process participation alone is not a completed permit or a guaranteed schedule.

What remains unknown Not established in this review

Final authorizations for the additions, tenant identity and commitments, detailed project budget, IT capacity, construction awards, utility agreements and final schedule remain unestablished by the initiation notice.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Meta Rosemount

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Rosemount links computing development to the cost of local services. Meta describes paying the full costs of its energy, water and wastewater service, which helps identify who is meant to bear the infrastructure burden. That commitment is economically different from utility profit or completed operation. The owner still needs useful computing outcomes sufficient to recover its capital and operating expenses.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Development described in operator fact sheet; opening not confirmed by reviewed evidence Readiness source →

What the sources establish

• Operator / company disclosure Investment scope. Meta’s fact sheet describes more than USD 800 million of data-center investment in Minnesota, about 100 operational jobs once completed and a future role in its global infrastructure. The document does not confirm a campus opening date, current IT load or final construction cost. Primary source →
• Operator / company disclosure Service and cooling commitments. The sheet identifies Xcel Energy planning and City of Rosemount water and wastewater coordination, with Meta paying full service costs. It describes a planned closed-loop liquid-cooling system using zero operational water. That cooling claim is not evidence of zero total site water use or measured operation. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: cost responsibility is a useful de-risking commitment for the community, while it can raise Meta’s all-in project cost. Utility shareholders benefit only under the actual asset, tariff and financing arrangements. Renewable matching and closed-loop design address different issues; neither demonstrates every hour of carbon-free operation or eliminates all water demand. Actual operating reports are needed to compare the design with delivered outcomes.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Delivery can depend on several infrastructure parties. Equipment or liquid-cooling upgrades may raise capital needs. Service costs can exceed early expectations, and resource design claims may not match actual operating conditions or workload density.

What would change our view?

Our analysis: Our confidence would rise with operating confirmation, transparent service terms and measured efficiency. It would fall if infrastructure obligations expand without matching computing benefits or if delivery remains unverified after substantial funding.

What remains unknown Not established in this review

Opening date, current IT MW, hardware suppliers, commissioned cooling configuration, actual water consumption, service tariffs and campus financial contribution remain unestablished.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

DataBank Red Oak campus

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Red Oak has stronger contractual and financing evidence than a speculative land announcement. The initial buildings are leased, and additional funding supports a fourth building. The investor question now shifts toward delivery, debt service and lease cash flow. A loan commitment reduces funding uncertainty but is an obligation to repay, not revenue or profit.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Construction financing expanded in June 2026; current commissioning unestablished Readiness source →

What the sources establish

• Operator / company disclosure Initial financed scope. DataBank’s April 21, 2026 disclosure reports a USD 2 billion construction loan for DFW9–DFW11, three leased buildings totaling 600,000 square feet and 180 MW of planned power. MUFG is named as administrative agent and lead arranger. Financing and leasing do not establish commissioning. Primary source →
• Operator / company disclosure Later financing update. The June 15, 2026 release adds USD 650 million for a fourth building with 60 MW of incremental IT capacity, bringing Red Oak financing to USD 2.65 billion. The upsize includes USD 400 million bank financing and USD 250 million private-placement notes. Primary source →
• Operator / company disclosure Corporate boundary. The June release separately announces an USD 800 million corporate revolver. It is general corporate funding and should not all be counted as additional Red Oak financing. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: preleasing reduces some demand uncertainty, but contract cancellation, delivery obligations and commencement timing still matter. Undrawn commitments differ from cash spent and debt outstanding. The later financing update changes the development funding scope, while the separate corporate revolver serves broader purposes. Returns to owners depend on cash left after lenders and ongoing capital needs, not on the size of the financing headline.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Late handover can defer income while interest accrues. Refinancing terms can worsen, and concentrated tenants can increase renewal risk. Loan covenants can constrain distributions even after operational success.

What would change our view?

Our analysis: Our confidence would rise with on-time handovers and contracted rent covering full debt service. It would fall with cost overruns, delayed billing or refinancing needs that absorb the anticipated equity cash.

What remains unknown Not established in this review

Current commissioned load, tenant identity, lease price and duration, loan rates and maturities, drawn balances, final cost and site cash flow remain unestablished by the reviewed releases.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

CyrusOne Allen: separate operating capacity from landbank optionality

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: Allen combines an existing capacity story with a future landbank option. The operator explicitly distinguishes DFW3 and DFW4 from DFW5. Treating full-buildout floor area or future land as current earnings would overstate the investment case; existing leasing and incremental expansion returns need separate models.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. DFW3 and DFW4 listed facilities; DFW5 identified as landbank Readiness source →

What the sources establish

• Operator / company disclosure Present versus future scope. CyrusOne identifies DFW3 and DFW4 as the campus facilities providing 77 MW of IT capacity, while DFW5 represents landbank for future development. Primary source →
• Operator / company disclosure Full-buildout area. The operator describes 976,000 square feet of technical space at full buildout across a campus exceeding 90 acres. The area is not evidence that every planned building is delivered. Primary source →
• Operator / company disclosure Cooling specifications. The listed technical specifications include closed-loop air-cooled chilled water and cooling capacity of 150–180 watts per square foot. These do not establish acceptance of a particular high-density AI rack design. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: existing infrastructure may support incremental deployment more efficiently than an entirely new campus, but spare capacity and usable power must be demonstrated. Landbank value comes from the ability to choose when to invest, not from recording unbuilt revenue. Underwrite recurring cash from existing facilities separately from the cost and execution risk of DFW5. Floor-area figures are especially weak revenue proxies when customers differ in power density and cooling requirements.

Milestones that would change the assessment

What could break the investment case?

Our analysis: A campus-wide headline can blend operating assets with unfinished options. High-density customers may require capital beyond the listed cooling design, and cost recovery depends on lease terms. Existing customers can renew at different densities or rates, making aggregate square footage a poor indicator of future cash generation.

What would change our view?

Our analysis: Stable existing-facility leasing and an expansion backed by enforceable demand and power would strengthen the case. Idle capacity, owner-funded upgrades without adequate rent compensation or speculative DFW5 spending would weaken it. No value is assigned to a future building merely because land is available.

What remains unknown Not established in this review

Current billable occupancy, delivered versus available MW, DFW5 schedule and power allocation, customer names, rent rates, project budgets, measured PUE and equipment-specific AI suitability are not disclosed in the cited campus overview.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Sabey SDC Seattle: distinguish vacancy from future inventory

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: Sabey’s announcement combines then-available operating inventory with future preleasing inventory. Those quantities answer different questions. Filling existing space can generate near-term receipts, while delivering expansion capacity requires additional capital and execution. Both need updated evidence.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing campus; October 2025 release targeted expansion delivery in 2027 Readiness source →

What the sources establish

• Operator / company disclosure Dated availability. Sabey’s October 2025 inventory announcement offered 3.5 MW then available at SDC Seattle and 12 MW for preleasing, with the latter targeted to come online in 2027. Primary source →
• Operator / company disclosure Correct scope. The announcement lists multiple campuses and dates. Its Seattle quantities should not be combined with Columbia, Umatilla or other campuses, and “available for preleasing” does not establish signed leases. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: the near-term case depends on leasing inventory already delivered, while the expansion case depends on a forward development yield. Preleases may reduce demand risk yet leave construction, power and fit-out exposure with the owner. An inventory figure can shrink because customers signed, capacity was reserved, or plans changed; only an updated reconciliation identifies which explanation applies.

Milestones that would change the assessment

What could break the investment case?

Our analysis: An old availability announcement can be misread as today’s vacancy. Reservations can overstate demand if cancellation terms are weak. Expansion may compete with the existing campus for staff, power or customers, and new rent can require substantial enabling spending.

What would change our view?

Our analysis: Updated leasing and construction evidence showing orderly conversion to billing would strengthen the case. Uncommitted expansion spending, repeated target changes or unclear inventory reconciliation would weaken it. No delay is asserted merely because current progress is not disclosed here.

What remains unknown Not established in this review

Current vacancy, signed preleases, delivered expansion MW, budget, tenants, rental rates, utility arrangements and specific 2027 commissioning dates remain unestablished by the October 2025 update.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

DataBank SEA2: sell connectivity and usable density

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: SEA2 is a smaller colocation and connectivity asset with advertised density options. Its investment value should be tested through lease mix, cross-connect economics and the cost of supporting denser customers. Technical capability creates an opportunity; it does not prove AI equipment is installed or producing revenue.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing marketed facility; historical expansion described separately Readiness source →

What the sources establish

• Operator / company disclosure Current listed specifications. DataBank lists SEA2 at 3311 South 120th Place, with 28,240 IT square feet, 3 MW of critical IT capacity and 17 onsite carriers. These are operator-listed specifications, not measured utilization. Primary source →
• Operator / company disclosure Density offering. The page advertises 10 kW-plus air-cooled and 25 kW-plus liquid-cooled cabinet density, HPC capability, 2N power and N+1 cooling. Customer-specific acceptance still requires evidence. Primary source →
• Operator / company disclosure Avoid double counting. A separate expansion section on the same page describes adding 15,000 square feet and 2 MW. It does not establish that this should be added again to the listed 3 MW total. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: a dense customer can increase rent per cabinet while consuming a disproportionate share of electrical and cooling capacity. Compare contribution per constrained kW rather than simply cabinet occupancy. Interconnection can support retention and service margin, but network presence must turn into paid connections. Expansion snippets and current specification tables need reconciliation before deriving any capacity growth rate.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Advertised density may depend on the specific room, equipment and cooling arrangement. A high-power tenant can exhaust available capacity while many cabinets remain empty. Undated expansion language can also create an inflated total if added to current specifications.

What would change our view?

Our analysis: Documented dense deployments with appropriate pricing and limited unrecovered capital would strengthen the case. Rising support costs, power exhaustion or expensive retrofits without compensated contracts would weaken it. An actual named AI workload would improve attribution beyond general HPC capability.

What remains unknown Not established in this review

Current billable capacity, workload mix, density approvals, liquid-cooling installation scope, expansion commissioning dates, price, maintenance capital, measured efficiency and tenant identities remain undisclosed in the listing.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

DataBank BOS1: a connectivity business at modest power scale

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

Our analysis: BOS1’s small electrical footprint makes connectivity and service quality more relevant than a hyperscale training narrative. Investors should ask whether its location and network access support durable customer relationships and profitable service receipts. A fiber connection is a documented technical relationship, not a quantified contract.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Existing colocation and interconnection facility Readiness source →

What the sources establish

• Operator / company disclosure Facility scale. DataBank lists BOS1 at 1 Summer Street, Suite 704, with 7,760 IT square feet, 0.5 MW of critical IT capacity and five onsite carriers. Primary source →
• Operator / company disclosure Connectivity role. The operator describes a carrier-hotel location, a direct Zayo fiber feed and partner cloud connectivity. It does not name a dedicated AI tenant or training cluster. Primary source →
• Operator / company disclosure Technical offering. The listing specifies 2N power, N+2 cooling and air-cooled cabinet density of 10 kW-plus. These are product specifications rather than operating-load measurements. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: proximity to networks can be more economically important than sheer megawatts. A small deployment may pay for low-latency access, resilience and convenient connections rather than maximum compute scale. Evaluate the physical room’s economics separately from the wider carrier hotel. Building-wide network claims or market demand do not automatically accrue to DataBank’s suite. Density upgrades also need a credible return within a limited power envelope.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Limited power can constrain compute-heavy growth. Customers may value the broader building more than the specific operator suite, increasing competitive pressure. A direct fiber feed establishes connectivity but not exclusivity, pricing power or a material revenue stream.

What would change our view?

Our analysis: Stable retention, profitable connection growth and disciplined upgrade spending would strengthen the case. Customer migration, weak service attach rates or capital-intensive modifications within a tight power envelope would weaken it. AI relevance should rest on actual workload evidence rather than downtown proximity.

What remains unknown Not established in this review

Occupancy, billable power, paid connections, Zayo contract economics, tenant identities, workload mix, site income, property costs and upgrade budgets remain undisclosed in the cited facility listing.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

AWS Mexico (Central)

Compare this project → · Track milestones → · How to read the evidence labels

Expanded analysis · Original disclosures linked below · CAD and USD preserved as labeled · Company statements attributed; interpretations labeled.

The one-minute investment takeaway

AWS Mexico is an operating cloud-region story. The central question is whether local infrastructure attracts additional paid usage or mainly relocates workloads already billed elsewhere in AWS. Residency and proximity can strengthen the offering, but moving an existing customer between regions does not automatically create incremental Amazon revenue.

Readiness basis: Classification of evidence in the reviewed profile; strongest documented phase, not full-campus completion. Region launched January 14, 2025; regional economics undisclosed Readiness source →

What the sources establish

• Operator / company disclosure Operating evidence. AWS announced general availability of mx-central-1 on January 14, 2025 with three Availability Zones. The launch disclosure describes local data storage, lower latency and cloud, AI and machine-learning capabilities. Three zones do not mean three buildings. Primary source →
• Operator / company disclosure Investment boundary. The same announcement sets out more than USD 5 billion of investment in Mexico over 15 years. This is a country-level commitment over time, not current-year expenditure, the cost of one campus or forecast cloud revenue. Primary source →

Who receives the money?

Our analysis of disclosed roles. Revenue routes are business-model interpretations, not estimates of booked project revenue.

How financially meaningful is it? Analyst interpretation

Our analysis: the useful unit is paid service consumption rather than the investment headline. New local workloads may add revenue; relocating existing workloads may instead improve retention, latency or compliance. Those benefits can matter without appearing as new regional sales. The investment commitment cannot be divided by 15 and treated as an annual spending schedule because timing and scope are undisclosed.

Milestones that would change the assessment

What could break the investment case?

Our analysis: Customer migrations can inflate apparent growth without adding total sales. Price competition, power costs and slow utilization can dilute returns. Region availability alone does not prove every AI product or accelerator is offered locally.

What would change our view?

Our analysis: Our confidence would rise with documented new workloads and sustained customer expansion. It would fall if adoption largely consists of transfers, if service gaps prevent deployments, or if capacity grows faster than paid demand.

What remains unknown Not established in this review

Regional revenue, operating MW, utilization, AI-specific usage, project margins and parcel-level facilities are not disclosed in the reviewed announcement. No regional earnings forecast is justified.

Project research for information, not a buy recommendation. Financial illustrations use assumptions, not disclosed site results or management guidance.

Permanent project profile →

Microsoft Mexico Central

Compare this project → · Track milestones → · How to read the evidence labels