Financial evidence, stock performance and the questions that connect this company to the physical AI buildout.
Our analysis: documented hardware participation supports an infrastructure-demand story. The investment assessment still depends on company-wide sales, margins, customer concentration and the cost of product transitions—not the sum of campus announcements.
Existing financial snapshots retain their original dates below. This editorial synthesis does not refresh financial data or provide a valuation recommendation.
Economic role: Named GPU platform; order value and earnings contribution unverified
Readiness and scope: Under construction · 44 MW powered-capacity plan; 13,800 GB300 GPUs financed, not verified installed load
Financing disclosed and construction reported in March; subsequent operational acceptance not verified · Evidence dates and conditions are retained in the full project profile.
Economic role: Named GPU platform; order value and earnings contribution unverified
Readiness and scope: Operating · 26 MW SIN01 reported by DCD; 1.2 GW permitted campus envelope; second 200 MW building planned
SIN01 operating; larger Rubin deployment and second-building capacity remain future scopes · Evidence dates and conditions are retained in the full project profile.
Economic role: Named GPU platform; order value and earnings contribution unverified
Readiness and scope: Under construction · 230 MW campus plan; 100 MW contractor building package; neither is verified operating IT load
Contractor reports works underway September 2026; Microsoft agreement supersedes historical OpenAI scope · Evidence dates and conditions are retained in the full project profile.
Economic role: UF identifies installed DGX B200 systems and collaboration; project contract value and NVIDIA revenue recognition unknown
Readiness and scope: Operating · UF reports 63 DGX B200 nodes / 504 GPUs; campus IT MW and comparable utility load unknown
HiPerGator fourth-generation launch and installed system documented in October 2025; operating MW unverified · Evidence dates and conditions are retained in the full project profile.
Economic role: Operator identifies Blackwell computing/networking deployment; procurement value and earnings recognition unverified.
Readiness and scope: Operating · Reported 8 MW colocation agreement; operating IT MW unverified. Government program allocation: 1,000 B200 accelerators; wider commercial deployment separate.
AI computing operation confirmed by authority January 18, 2026; current voucher program active; measured utilization unknown · Evidence dates and conditions are retained in the full project profile.
Economic role: Referenced technology supplier in Nebius AI-factory strategy; final campus inventory and supplier revenue unverified
Readiness and scope: Under construction · Up to 310 MW planned full development; current energized IT load unverified
Construction reported underway; first customer capacity targeted for 2027 · Evidence dates and conditions are retained in the full project profile.
Economic role: Reports Blackwell wafer volume production in Phoenix; Arizona production allocation and procurement value undisclosed
Readiness and scope: Operating · First fab N4 production; later N3 and N2/A16 phases; wafer starts and operating packaging throughput not established in this review
First fab in volume production; later fabs and packaging have separate construction and planned phases · Evidence dates and conditions are retained in the full project profile.
customer deployment delays, concentration, hardware transitions, competitive pricing and export restrictions can affect demand and economics. A campus naming NVIDIA does not establish an order value or recognized revenue.
Monitor disclosed financial results and margins, documented product delivery and customer commissioning. Stronger verified demand and cash conversion would strengthen the assessment; delays, weaker margins or concentration pressure would weaken it.
Project-specific order values, revenue recognition, customer utilization and earnings attribution are generally not established in these campus records. Data Center segment revenue is not a comparable AI-only measure.
Power and the AI Buildout → · Catalyst tracker → · Weekly Investor Briefing →
Issuer identifies 44 MW powered capacity, 13,800 GB300 GPUs and European 200 MW ambition; none is proof of installed utilization.
Two funding tranches of approximately USD 720m and EUR 94m; identifies the seven-bank syndicate and GPU procurement purpose.
Independent reporting of USD 830m debt and the original Q2 2026 operating target; not current commissioning confirmation.
Reports construction launched and planned service in Q2 2026. Supports a dated construction classification, not October operation.
Reports SIN01’s first full operating-year performance; annual renewable matching and measured water figures distinguish operations from targets.
Announces Microsoft agreement, 66,000+ Rubin GPUs starting late 2027, EUR 230m shared infrastructure and EUR 465m second building.
Independent reporting of 26 MW SIN01, expanded compute agreement, second-building investment and future deployment schedule.
Identifies Microsoft agreement and 30,000+ Rubin GPUs in 2027 at the 230 MW campus.
Identifies USD 790m committed financing and separate USD 790m uncommitted accordion for another 115 MW.
Reports three North Campus buildings totaling 100 MW, works underway and 2027 completion expectation.
Independent reporting of financing and Microsoft capacity relationship replacing the earlier OpenAI plan.
Establishes Aker partnership, hydropower rationale and historical Stargate/OpenAI plan; does not verify the current customer contract.
LFIR 3066 requests $24 million for University of Florida AI Leadership Technology Renewal / HiPerGator Updates, through the Board of Governors, including NVIDIA networking. This project-specific state record is a funding request, not an enacted appropriation, cash receipt or final contract.
Public university disclosure records the East Campus launch and reports a $33 million purchase, 63 NVIDIA DGX B200 nodes and 504 GPUs replacing the earlier A100 SuperPOD. Performance comparisons are issuer claims and not comparable campus MW. The purchase figure differs in scope/date from the earlier $24 million budget/request.
Original on-site journalism documents the October 14 ribbon cutting and media tour, the East Campus location and the December 2024 $24 million trustee allocation. Energy-efficiency assertions are attributed to directors; it does not establish a power tariff or measured IT load.
Original reporting and photographs document the launch and guided data-center tour. It distinguishes the earlier $24 million trustee investment from the computing generations and attributes NVIDIA installation involvement to UF. Research-budget figures discussed at the event are not project revenue.
Project-specific government authority confirms Nebius national-supercomputer infrastructure is live and operational, opening subsidized-access allocation. Describes 1,000 B200 accelerators and 70% industry / 30% research allocation. This confirms computing service, not a comprehensive site permit or measured utilization.
Government tender-award disclosure selects Nebius for national computing infrastructure, with projected investment exceeding NIS500 million including NIS160 million government support. Award and support amounts are not proof of cash disbursement or incurred expenditure. English/Hebrew versions are one underlying document.
Current official voucher call identifies Nebius as operator and a 1,000-B200 monthly allocation framework; 30% industry and 80% academic price discounts subject to eligibility and committee approval. Open date January 1 is not publication date. Detailed future application windows are not actual utilization or recipient awards.
Operator announces AI Cloud deployment at a Modiin colocation facility using NVIDIA Blackwell infrastructure. Its 4,000 HGX B200 wording describes the wider deployment and is not the government’s 1,000-accelerator allocation or audited present hardware utilization. Location and deployment claims attributed to company.
Original business reporting, including Nebius confirmation and attempted Mega Or inquiry, identifies existing 8 MW Modi’in site hosting the national supercomputer. Describes wider expansion separately. Capacity is reported commercial scope, not measured IT load; future additional 16 MW is not current delivered load.
Original interview with authority CEO supports competitive award, Blackwell plan and projected funding. Historical early2026 operating forecast is superseded by January 2026 authority confirmation. Interview is independent journalism, but award figures originate with authority, not an independent spending audit.
Business reporting describes Mega DC’s earlier 8 MW Modi’in agreement and reported September operation. Expected NIS35 million annual facility NOI is a company forecast; separate Masmiyya/Beit Shemesh 80 MW deal, $880 million construction estimate and NIS300 million NOI do not belong to this campus.
March 31 announcement: up to 310 MW planned full campus; first customer capacity targeted for 2027.
Municipal record: Pajarila location, city land lease, Polarnode development role and Nebius operator role.
August 11 reporting: building work began July 2026; Fira first-phase main contractor.
September follow-up: site clearing began March 2026.
Municipal FAQ: site and infrastructure context; participant-attributed answers are labeled.
May 13, 2026 contractor disclosure: approximately EUR 15 million substation agreement; 70 MW first phase and spring 2027 completion target. These are electrical infrastructure specifications and expectations, not verified energized IT capacity or customer service.
Arizona campus disclosure: first fab N4 high-volume production began Q4 2024; second fab N3 production targeted second-half 2027; third fab N2/A16 planned by decade end; fourth fab and first packaging plant initial construction begun in early 2026. USD 265 billion is an announced investment envelope.
July 16 announcement describes another USD 100 billion, bringing the announced Arizona plan to USD 265 billion and 10 fabs, two packaging facilities and R&D. This is announced scope, not delivered facilities, paid spending or a blanket construction permit.
Documents Fab 3 topping-out ceremony May 5. A highest-beam building milestone is not tool installation, production qualification or operating acceptance.
Reports additional Arizona investment and management’s market-dependent expansion timing; corroborates the announcement, not completion.
Reports Blackwell wafer volume production at TSMC Phoenix. Wafer production is not evidence of all-US packaging or Arizona revenue allocation.
Up to USD 6.6 billion direct funding for the historical more-than-USD-65-billion three-fab scope; milestone-based disbursement. Up to USD 5 billion proposed loans; not proof of full disbursement or funding for all later expansion.
File 26-0966, Ordinance S-52899: adopted wastewater-development authorization, infrastructure costs assigned to TSMC, reclamation conditions and June 30, 2028 initial-plant deadline in the summary. Adoption does not prove execution of agreement or completed works.
Lists TSMC Phoenix Project, facility F041443, 5088 Innovation Circle, as Active. Index is not a complete permit-condition, amendment or compliance audit.
10-year Arizona packaging/testing procurement collaboration; partner campus distinct from TSMC fabs and its own packaging facility. Contract allocation and realized revenue not disclosed.
USD 2 billion initial five-year TSMC interposer manufacturing agreement, Malta New York, volume ramp targeted first-half 2028. Supply-chain context, not a Phoenix fab or verified Arizona allocation.
Reports being on track to purchase well over 100 million TSMC Arizona chips in 2026. This is a sourcing expectation, not audited final annual purchases or an AI-only workload allocation.
Original USD 12 billion plan and 20,000-wafer/month target, not current measured output.
Two-fab USD 40 billion historical scope. Earlier targets are not substituted for current production evidence.
USD 165 billion expansion envelope with further fabs, packaging and R&D; issuer-authored filing, not independent corroboration.
Facilities, equipment and process technician training with education partners. Programme existence does not establish workforce sufficiency or production ramp success.
Core/shell versus emissions-unit construction correspondence. Applicant positions are separated from EPA response; not a campus-wide operating permit.
Reports executive goal for Arizona CoWoS/3D-IC capability before 2029. Target, not delivered capability.
Reports USD 265 billion planned investment and total 12 manufacturing/packaging facilities; not 12 operating wafer fabs.
October 15, 2026 earnings conference is the next company disclosure checkpoint; future results are not known.
Original memorandum describes planned Peoria packaging/test collaboration for TSMC Phoenix customers; no completed partner capability established.
2025 annual filing confirms first Arizona fab high-volume production at end-2024, additional land acquired in 2026, and conditions, compliance and project milestones attached to CHIPS funding.
Management discusses Arizona expansion, global USD 60–64 billion 2026 capital budget and collective overseas-fab gross-margin dilution. Guidance is not site-level expenditure or Arizona-only profitability.
Board approved second-quarter financial statements and approximately USD 29.4425 billion capital appropriations for advanced technology, packaging and fab facilities. Appropriations are companywide authorizations, not Arizona expenditure or an amount to add to annual capital guidance.
Total annual company revenue · year ended 2026-01-25. This is company-wide revenue.
Annual financial evidence ↗ · Compare the revenue rankingFor fiscal Q2 2027, ended July 26, NVIDIA reported $96.2 billion in revenue, up 106% year over year. Data Center revenue was $89.0 billion, up 117%. Data Center is a broader segment; it is not a separately comparable AI-only revenue measure.
Read the primary announcement ↗Selected dated update; this editorial snapshot does not refresh automatically.
Data Center revenue is a separate financial signal. NVIDIA’s Data Center segment includes accelerated computing and networking. The entire segment is not classified here as AI-only revenue.
Disclosure and basis ↗Can customers keep deploying complete GPU systems at a pace that supports demand, while NVIDIA maintains margins? Watch data-center growth, export restrictions and the transition between product generations.
These are research questions, not a valuation conclusion or a recommendation to buy or sell.
These profiles mention the company. A mention does not establish a supplier contract or project-specific revenue.
These profiles contain a named role or connection. A relationship does not establish booked revenue, ownership of every project layer or an attractive valuation. Site-attributable financial materiality remains a separate diligence question.
University of Florida identifies installed NVIDIA DGX B200 systems at HiPerGator. Project contract value and NVIDIA revenue recognition remain unverified.
Read HiPerGator research →Beyond.pl’s F.I.N. uses NVIDIA DGX B200 hardware and networking, according to the operator’s September 2025 commissioning disclosure. The project role does not establish contract value or NVIDIA revenue.
Project evidence and delivery history · Read the analysis · Operator commissioning disclosure
Each connection below has a specific, cited role. A supplier, lender, customer or proposed acquirer has different economics from the project owner.
Crusoe reports Oracle delivered GB200 racks; hardware supplier relationship, not disclosed site procurement revenue.
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
Disclosed DGX B200 compute and networking technology supplier/technical collaborator. No project contract value or attributable revenue verified.
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
NVIDIA-powered Beta system named by hosting University; contract size not disclosed
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
GPU technology supplier: Microsoft describes NVIDIA GB200 systems as the computing architecture for its Wisconsin Fairwater AI campus.
The source describes the campus architecture and intended scale. It does not establish NVIDIA campus ownership, project-specific revenue or independently measured installed GPU inventory.
GPU technology supplier: Microsoft describes NVIDIA GB200 NVL72 systems in its Atlanta Fairwater design and connected AI-computing network.
Network-wide architecture and scaling claims must not be converted into an installed Atlanta GPU count, NVIDIA campus ownership or project-specific revenue.
NVIDIA GB300 GPU technology in IREN's Childress Horizon deployments for Microsoft.
The hardware relationship does not establish campus ownership or allocate the entire Dell purchase or Microsoft cloud contract to NVIDIA revenue. Later phases retain their stated delivery and acceptance limits.
NVIDIA GPU technology specified for the Prince George AI Cloud installation.
Specific disclosed equipment plan, not proof all GPUs or the hall are commissioned today; no NVIDIA site revenue inferred.
Technology supplier: transaction counsel describes financing for 13,800 NVIDIA GB300 GPUs for Mistral AI's facility.
Planned equipment acquisition does not disclose NVIDIA recognized revenue, delivery acceptance or ownership of the campus.
Blackwell hardware and networking identified by operator; B200 program allocation confirmed by authority.
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
NVIDIA Rubin is the announced future hardware platform; installed units and supplier revenue are unverified.
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
NVIDIA hardware is specified for the future expansion; installed quantity and procurement revenue remain unverified.
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
NVIDIA is a semiconductor customer whose Blackwell wafers are manufactured at TSMC's Phoenix facility.
Manufacturing evidence does not quantify site purchases or NVIDIA revenue, and does not establish that all U.S. packaging is performed there.
Installed DGX B200 technology supplier; contract revenue unknown
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.
Hopper GPU and SpectrumX networking supplier as of Oct 2024
The documented role does not establish contract value, project-level earnings, accepted operating load or a share of company revenue.