AI compute deals

Google agreed to fund uprates at 11 Constellation nuclear units in Illinois, Pennsylvania and New Jersey that add 890 MW to the PJM grid, under a 20-year deal that also includes a separate 15-year supply agreement for 2,700 MW from Constellation's existing PJM fleet; Constellation put its new investment at more than $4.3 billion.

Term: 20 years.

• Company Constellation Energy — Google and Constellation Announce Landmark Agreement to Bring 890 MW of New Nuclear Capacity to PJM Grid as Part of Long-Term Power Deal, 2026-10-06

CoreWeave entered India with a 240 MW AI cloud deployment at AdaniConneX data centers.

• Company CoreWeave — CoreWeave Enters India With 240 MW AdaniConneX Deployment, 2026-10-06

Reuters reported, citing Anthropic's confidential IPO filing, that Broadcom agreed to lend Anthropic up to $42 billion to lease its chips. Broadcom did not comment and Anthropic declined to comment.

Stated as “up to $42 billion”.

Related: Anthropic and Broadcom (2025-12-11).

• Press Reuters (via The Star) — Exclusive-Broadcom to lend Anthropic up to $42 billion to lease its chips, filing says, 2026-10-01

Akamai expanded its relationship with Anthropic to $11.6 billion of contractual commitment over seven years for Anthropic's CPU workloads on Akamai Cloud, with room to grow by up to $9 billion more. Akamai issued Anthropic a warrant for up to about 5% of its stock that vests as the commitment expands.

Term: 7 years. Stated as “$11.6 billion of contractual commitment over seven years”.

• Filing Akamai via SEC EDGAR — Akamai Announces $11.6 Billion expanded relationship with Anthropic (Exhibit 99.1), 2026-09-24
• Filing Akamai via SEC EDGAR — Akamai-Anthropic release (potential commitment), 2026-09-24

Reuters reported, citing a source, that Anthropic signed a $35 billion cloud deal with NVIDIA-backed Lambda for about 350 MW at a data center Hut 8 is developing in Nueces County, Texas. Neither company announced the deal.

With Hut 8, NVIDIA. Stated as “worth $35 billion”.

Related: Anthropic and Nscale (2026-08-26); Microsoft and Lambda (2025-11-03); Anthropic (reported) and Volta Infra (2026-08-04).

• Press Reuters (via The Star) — Anthropic signs $35 billion cloud deal with Nvidia-backed Lambda, source says, 2026-09-01
• Press Reuters (via The Star) — Reuters report (capacity), 2026-09-01

Press reports, led by Bloomberg, said Anthropic agreed to spend about $45 billion over six years renting about 460 MW of NVIDIA Vera Rubin capacity at Nscale's Monarch campus in West Virginia, a site Microsoft had earlier signed a letter of intent for and then left. The companies did not announce the customer or terms.

Chips: NVIDIA Vera Rubin. Term: 6 years. Stated as “$45B over six years”.

Related: Anthropic and Lambda (2026-08-31); Microsoft and Nscale (2025-10-15); Anthropic (reported) and Volta Infra (2026-08-04).

• Press The Next Web — Anthropic signs $45B compute deal with Britain's Nscale, 2026-08-26

SB Energy will build, own and operate an AI campus in Pike County, Ohio under a 20-year lease to OpenAI, starting with 4.25 GW of IT load and up to about 8 GW if NVIDIA exercises an option, and the site will host NVIDIA systems. NVIDIA provided lease guarantees capped at $105 billion for the initial 4.25 GW and agreed to invest $1.5 billion in SB Energy.

With NVIDIA, SoftBank Group. Term: 20 years. Stated as “NVIDIA's aggregate payment obligation is cumulatively capped at $105 billion for its initial commitment”.

Related: OpenAI and SB Energy (2026-01-09); OpenAI and NVIDIA (2025-09-22).

• Filing NVIDIA via SEC EDGAR — NVIDIA Form 8-K (August 17, 2026), 2026-08-17
• Filing NVIDIA via SEC EDGAR — NVIDIA, SB Energy and OpenAI PORTS-Pike campus press release (Exhibit 99.1), 2026-08-17

Volta Infra said it signed a $10 billion contract with an unnamed AI company for cloud capacity in Europe alongside Bitdeer; Bloomberg reported the customer is Anthropic and that the six-year deal covers a 133 MW Bitdeer data center in Norway with NVIDIA Vera Rubin chips. Anthropic declined to comment.

With Bitdeer. Chips: NVIDIA Vera Rubin. Term: 6 years. Stated as “a $10 billion contract”.

Related: Anthropic and Nscale (2026-08-26); Anthropic and Lambda (2026-08-31).

• Press Reuters (via The Star) — AI cloud startup Volta valued at $2.4 billion, announces $10 billion AI partnership, 2026-08-04
• Press Reuters (via The Star) — Reuters report (customer and capacity), 2026-08-04

Meta and BlackRock formed a venture, 80% owned by BlackRock funds and 20% by Meta, to develop and own Meta's 1 GW El Paso, Texas data center campus, with about $14 billion in total development costs. Meta leases the whole campus on a four-year initial term with options extending to a potential 20 years.

With Global Infrastructure Partners, HPS Investment Partners. Term: 4 years. Stated as “approximately $14 billion in total development costs”.

Related: Meta and Blue Owl Capital (2025-10-21).

• Company Meta — Meta Announces New Strategic Venture With BlackRock to Develop Data Center in El Paso, 2026-07-28
• Company Meta — Meta El Paso venture (capacity), 2026-07-28

AMD and Anthropic agreed that Anthropic will deploy up to 2 GW of AMD Instinct MI450 Series GPUs in Helios racks, with the first gigawatt starting in the first half of 2027. AMD committed to a strategic equity investment of up to $5 billion in Anthropic; no purchase value was disclosed.

Chips: AMD MI450, AMD MI455X. Starts 2027. Stated as “a strategic equity investment of up to $5 billion in Anthropic”.

Related: OpenAI and AMD (2025-10-06); Meta and AMD (2026-02-24).

• Company AMD — AMD and Anthropic Announce Strategic Partnership to Deploy up to 2 Gigawatts of AMD Instinct MI450 Series GPUs, 2026-07-22
• Company AMD — AMD-Anthropic release (investment), 2026-07-22

OpenAI announced Project Camellia, a data center it is designing and developing itself in Effingham County, Georgia, and said it is contracting with Georgia Power for 3.2 GW of power delivered in phases from 2028 to 2032.

Starts 2028.

• Company OpenAI — Building AI infrastructure with the Effingham County community, 2026-07-22

TeraWulf signed a 20-year lease with Anthropic for a purpose-built AI campus of about 401 MW of critical IT load at its Justified Data site in Hawesville, Kentucky, expected to generate about $19 billion over the initial term, with first capacity in the second half of 2027. TeraWulf also agreed to sell its stake in the Abernathy, Texas joint venture to a Fluidstack-led group.

Term: 20 years. Starts 2027. Stated as “approximately $19 billion of contracted revenue over the initial lease term”.

Related: Anthropic and Hut 8 (2025-12-17); Anthropic and Fluidstack (2025-11-12).

• Filing TeraWulf via SEC EDGAR — TeraWulf Announces Anthropic Lease at Justified Data (Exhibit 99.1), 2026-07-06
• Filing TeraWulf via SEC EDGAR — TeraWulf release (capacity), 2026-07-06

SK Group and NVIDIA signed letters of intent for a partnership described as worth more than $500 billion, covering SK Telecom's planned 2 GW AI cloud in South Korea on NVIDIA's DSX AI factory platform and Vera Rubin systems, and long-term SK hynix HBM supply and co-development. The figure is an aggregate scope, not a disclosed purchase commitment; the first AI factory is expected in 2027.

With SK Telecom, SK hynix. Chips: NVIDIA Vera Rubin. Starts 2027. Stated as “a strategic partnership worth more than $500 billion”.

Related: OpenAI and Samsung Electronics (2025-10-01).

• Press eeNews Europe — SK Group and NVIDIA expand AI factory and HBM partnership, 2026-07-27
• Press eeNews Europe — SK Group and NVIDIA partnership (letters of intent), 2026-07-27

Sharon AI signed a six-year Master Cloud Services Agreement with NVIDIA, worth up to $4.88 billion, to build 72 MW of new AI data center capacity in Australia with up to 40,000 GB300 GPUs under a revenue-sharing and credit-support model. It reported the deal as $4.9 billion of its roughly $8.8 billion total contract value in August 2026.

Chips: NVIDIA GB300. Term: 6 years. Stated as “a contract value of up to $4.88 billion”.

Related: NVIDIA and IREN (2026-05-07); NVIDIA and Lambda (2025-09).

• Filing Sharon AI via SEC EDGAR — SharonAI Holdings Form S-1, 2026-07-31
• Filing Sharon AI via SEC EDGAR — Sharon AI Reports Second Quarter 2026 Results (Exhibit 99.1), 2026-08-06

SpaceX disclosed a Cloud Service Agreement under which Google pays $920 million per month from October 2026 through June 2029 for compute including about 110,000 NVIDIA GPUs, with a reduced fee during the ramp. Google may terminate or take fewer GPUs if SpaceX misses the September 30, 2026 delivery target, and either party may terminate on 90 days' notice after 2026.

With xAI. Starts 2026-10. Stated as “$920 million per month from October 2026 through June 2029”.

Related: Anthropic and SpaceX (2026-05-06); xAI and NVIDIA (2024-10-28).

• Filing SpaceX via SEC EDGAR — SpaceX free writing prospectus on Cloud Service Agreement with Google, 2026-06-05
• Filing SpaceX via SEC EDGAR — SpaceX FWP (capacity), 2026-06-05

IREN signed a five-year contract to provide NVIDIA with managed GPU cloud services for NVIDIA's internal AI and research work, worth about $3.4 billion, using air-cooled Blackwell systems in about 60 MW of its Childress, Texas data centers.

With Mirantis. Chips: NVIDIA Blackwell. Term: 5 years. Stated as “approximately $3.4 billion”.

Related: Microsoft and IREN (2025-11-03); NVIDIA and Lambda (2025-09); NVIDIA and Sharon AI (2026-06-08); IREN and NVIDIA (2026-05-07).

• Company IREN — IREN Secures $3.4bn AI Cloud Contract with NVIDIA, 2026-05-07

NVIDIA and IREN announced plans to deploy up to 5 GW of NVIDIA DSX-aligned AI infrastructure across IREN's sites, with its 2 GW Sweetwater campus in Texas as the flagship, and NVIDIA took a five-year right to buy up to 30 million IREN shares at $70 each, about $2.1 billion.

Stated as “a five-year right to buy up to 30 million IREN shares at $70 each”.

Related: NVIDIA and IREN (2026-05-07).

• Company NVIDIA — NVIDIA and IREN Announce Strategic Partnership to Accelerate Deployment of up to 5 Gigawatts of AI Infrastructure, 2026-05-07

Anthropic signed an agreement to use all of the compute capacity at SpaceX's Colossus 1 data center, more than 300 MW and over 220,000 NVIDIA GPUs. SpaceX's IPO filings said its Cloud Services Agreements with Anthropic cover about 325,000 NVIDIA GPUs across Colossus and Colossus II at $1.25 billion per month through May 2029, terminable on 90 days' notice after an initial three months.

With xAI. Chips: NVIDIA H100, NVIDIA H200, NVIDIA GB200. Starts 2026-05. Stated as “$1.25 billion per month through May 2029”.

• 2026-05-20: SpaceX's S-1 said the agreements cover Colossus and Colossus II at $1.25 billion per month through May 2029, terminable on 90 days' notice. Source

Related: xAI and NVIDIA (2024-10-28); Google and SpaceX (2026-06-05).

• Company Anthropic — Higher usage limits for Claude and a compute deal with SpaceX, 2026-05-06
• Press Tom's Hardware — Elon Musk's SpaceXAI to add another 660,000 AI GPUs this year, 2026-09-25
• Filing SpaceX via SEC EDGAR — Space Exploration Technologies Corp. Form S-1, 2026-05-20
• Filing SpaceX via SEC EDGAR — SpaceX final prospectus (Form 424B4), 2026-06-12

Nscale said it would deliver more than 66,000 NVIDIA Rubin GPUs to Microsoft at Start Campus's site in Sines, Portugal, investing EUR 230 million in shared infrastructure and EUR 465 million in a second 200 MW building.

Chips: NVIDIA Rubin.

Related: Microsoft and Nscale (2025-10-15); Microsoft and Nscale (2026-04-14).

• Company Nscale — Nscale to Deliver 66,000+ NVIDIA Rubin GPUs to Microsoft at Start Campus' Site in Portugal, 2026-05-05

Anthropic committed more than $100 billion over ten years to AWS technologies, securing up to 5 GW of new capacity across Graviton and Trainium2 through Trainium4, with nearly 1 GW of Trainium2 and Trainium3 capacity due by the end of 2026. Amazon invested $5 billion in Anthropic with up to $20 billion more possible.

With Amazon. Chips: AWS Trainium2, AWS Trainium3, AWS Trainium4, AWS Graviton. Term: 10 years. Starts 2026. Stated as “more than $100 billion over the next ten years to AWS technologies”.

Related: Anthropic and Amazon Web Services (2023-09-25); Anthropic and Amazon Web Services (2024-11-22); Anthropic and Amazon Web Services (2024-12-03).

• Company Anthropic — Anthropic and Amazon expand collaboration for up to 5 gigawatts of new compute, 2026-04-20

Nscale expanded its Microsoft agreement to add more than 30,000 NVIDIA Rubin GPUs at its 230 MW Narvik, Norway campus for delivery in 2027, after rolling up the former Aker-Nscale joint venture.

Chips: NVIDIA Vera Rubin. Starts 2027.

Related: Microsoft and Nscale (2025-10-15); OpenAI and Nscale (2025-07-31); Microsoft and Nscale (2026-05-05).

• Company Nscale — Nscale announces expanded deal with Microsoft in Norway, 2026-04-14

CoreWeave signed a multi-year agreement to run Anthropic's Claude workloads at production scale on its cloud, starting with a phased rollout that could expand over time. Financial terms were not disclosed.

Starts 2026.

Related: Meta and CoreWeave (2026-04-09).

• Company CoreWeave — CoreWeave Announces Multi-Year Agreement With Anthropic, 2026-04-10

CoreWeave and Meta signed an expanded agreement for AI cloud capacity through December 2032 worth about $21 billion, including some of the first NVIDIA Vera Rubin deployments. The total includes Meta exercising an option under its September 2025 order form.

Chips: NVIDIA Vera Rubin. Stated as “approximately $21 billion”.

Related: Meta and CoreWeave (2025-09-30).

• Filing CoreWeave via SEC EDGAR — CoreWeave and Meta Announce $21 Billion Expanded AI Infrastructure Agreement (Exhibit 99.1), 2026-04-09
• Filing CoreWeave via SEC EDGAR — CoreWeave Form 8-K (order form dated March 31, 2026), 2026-04-09

Anthropic signed an agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity coming online from 2027, mostly in the US, and later described it as a 5 GW agreement. The Information reported in May 2026 that Anthropic committed to spend about $200 billion with Google Cloud over five years; neither company confirmed that figure.

With Broadcom. Chips: Google TPU. Term: 5 years. Starts 2027. Stated as “US$200 billion (HK$1.56 trillion) with Google Cloud over five years”.

• 2026-05-05: The Information reported a roughly $200 billion five-year Google Cloud commitment by Anthropic and that Alphabet is investing up to $40 billion in Anthropic. Source

Related: Anthropic and Google Cloud (2025-10-23); Anthropic and Broadcom (2025-12-11); Anthropic and Fluidstack (2025-11-12).

• Company Anthropic — Anthropic expands partnership with Google and Broadcom for multiple gigawatts of next-generation compute, 2026-04-06
• Company Anthropic — Higher usage limits for Claude and a compute deal with SpaceX, 2026-05-06
• Press Reuters (via The Standard) — Anthropic commits to spending US$200 billion on Google's cloud and chips, the Information reports, 2026-05-06

Nebius signed a five-year agreement to provide Meta $12 billion of dedicated capacity on NVIDIA Vera Rubin systems from early 2027, and Meta committed to buy up to $15 billion of additional capacity on upcoming Nebius clusters, for a total contract value of up to about $27 billion.

Chips: NVIDIA Vera Rubin. Term: 5 years. Starts 2027. Stated as “a contract value of up to approximately $27 billion”.

Related: Meta and Nebius (2025-11-11); Nebius and NVIDIA (2026-03-11).

• Filing Nebius via SEC EDGAR — Nebius signs new AI infrastructure agreement with Meta (Exhibit 99.1), 2026-03-16
• Filing Nebius via SEC EDGAR — Nebius-Meta release (dedicated capacity), 2026-03-16

NVIDIA agreed to invest $2 billion in Nebius as part of a strategic partnership to help Nebius deploy more than 5 GW of capacity by the end of 2030 on NVIDIA's latest platforms.

Stated as “NVIDIA will invest $2 billion in Nebius”.

Related: Meta and Nebius (2026-03-16); CoreWeave and NVIDIA (2026-01-26).

• Filing Nebius via SEC EDGAR — NVIDIA and Nebius strategic partnership (Exhibit 99.1), 2026-03-11

Thinking Machines Lab said it struck a multi-year partnership with NVIDIA to procure at least 1 GW of Vera Rubin systems from early 2027, mainly for training, alongside a significant NVIDIA investment. Financial terms were not disclosed.

Chips: NVIDIA Vera Rubin. Starts 2027.

• Press Reuters (via Yahoo Finance) — AI startup Thinking Machines clinches capital and a major chip supply deal from Nvidia, 2026-03-10

OpenAI and AWS expanded their $38 billion agreement by $100 billion over eight years, with OpenAI committing to consume about 2 GW of Trainium3 and Trainium4 capacity. Amazon also agreed to invest $50 billion in OpenAI, starting with $15 billion and adding $35 billion when certain conditions are met.

With Amazon. Chips: AWS Trainium3, AWS Trainium4. Term: 8 years. Stated as “expanding their existing $38 billion multi-year agreement by $100 billion over 8 years”.

Related: OpenAI and Amazon Web Services (2025-11-03).

• Company OpenAI — OpenAI and Amazon announce strategic partnership, 2026-02-27
• Company OpenAI — OpenAI and Amazon announce strategic partnership (Trainium), 2026-02-27
• Company Amazon — Amazon's $50 billion investment in OpenAI: What to know, 2026-02-27

The Information reported that Meta agreed to a multibillion-dollar deal to rent Google Cloud TPUs for developing its AI models. Neither company announced terms.

Chips: Google TPU. Stated as “multibillion-dollar deal”.

Related: Meta and Google Cloud (2025-08-21).

• Press SiliconANGLE (citing The Information) — Google and Meta reportedly strike new, multibillion-dollar AI chip deal, 2026-02-26

AMD and Meta signed a definitive multi-year agreement for Meta to deploy up to 6 GW of AMD Instinct GPUs, starting with shipments of a custom MI450-based GPU in Helios racks in the second half of 2026. AMD issued Meta a warrant for up to 160 million AMD shares that vests as shipments scale; no dollar value was disclosed.

Chips: AMD MI450. Starts 2026. Stated as “performance-based warrant for up to 160 million shares of AMD common stock”.

Related: Meta and NVIDIA (2026-02-17); OpenAI and AMD (2025-10-06).

• Company AMD — AMD, Meta Announce Expanded Strategic Partnership to Deploy 6 Gigawatts of AMD GPUs, 2026-02-24
• Company AMD — AMD-Meta release (warrant), 2026-02-24

OpenAI said it will become the first customer of TCS's HyperVault data center business in India under its Stargate initiative, starting with 100 MW of capacity with potential to scale to 1 GW over time. No financial terms were disclosed.

With Tata Group.

Related: OpenAI and Stargate Project (2025-01-21).

• Press Outlook Business — OpenAI Becomes First Customer of TCS Data Centre Business, Aims 1 GW Capacity Building, 2026-02-19

NVIDIA and Meta announced a multiyear, multigenerational partnership covering millions of NVIDIA Blackwell and Rubin GPUs, standalone Grace CPUs and Spectrum-X Ethernet switches across Meta's own data centers and NVIDIA cloud partners. No financial terms were disclosed.

Chips: NVIDIA Blackwell, NVIDIA Rubin, NVIDIA Grace.

Related: Meta and AMD (2026-02-24).

• Company NVIDIA — Meta's AI Roadmap Supported by Large-Scale Deployment of NVIDIA CPUs, Networking and Millions of NVIDIA Blackwell and Rubin GPUs, 2026-02-17

NVIDIA bought $2 billion of CoreWeave stock at $87.20 per share and the two companies expanded their collaboration to help CoreWeave build more than 5 GW of AI factories by 2030 on NVIDIA platforms.

Stated as “NVIDIA has invested $2 billion in CoreWeave Class A common stock”.

Related: NVIDIA and CoreWeave (2025-09-15).

• Filing CoreWeave and NVIDIA via SEC EDGAR — NVIDIA and CoreWeave Strengthen Collaboration to Accelerate Buildout of AI Factories (Exhibit 99.1), 2026-01-26

OpenAI agreed to buy up to 750 MW of low-latency inference compute from Cerebras over three years, delivered from Cerebras-built or leased data centers in tranches through 2028. Reuters first reported the deal at more than $10 billion; Cerebras's IPO prospectus of May 2026 values it at more than $20 billion.

Term: 3 years. Stated as “more than $20 billion”.

• Company OpenAI — OpenAI partners with Cerebras, 2026-01-14
• Press Reuters (via Yahoo Finance) — OpenAI signs $10 billion computing deal with Nvidia challenger Cerebras, 2026-01-14
• Filing Cerebras Systems (SEC filing) — Cerebras Systems registration statement (Form S-1/A), 2026-05-04

Meta announced nuclear agreements with Vistra, TerraPower and Oklo supporting up to 6.6 GW of new and existing capacity by 2035, including 20-year Vistra PPAs for 2,176 MW, funding for up to eight TerraPower Natrium reactors and a 1.2 GW Oklo campus in Pike County, Ohio. The power supports grids serving Meta's Prometheus cluster in Ohio.

With TerraPower, Oklo.

• Press World Nuclear News — Meta announces 'landmark' agreements for new nuclear, 2026-01-09

OpenAI and SoftBank Group each invested $500 million in SB Energy, and OpenAI signed a 1.2 GW data center lease with SB Energy, which will build and operate OpenAI's Stargate site in Milam County, Texas.

With SoftBank Group. Stated as “OpenAI and SoftBank Group are each investing $500 million into SB Energy”.

Related: OpenAI and Oracle (2025-09-23); OpenAI and Stargate Project (2025-01-21).

• Company SB Energy — OpenAI and SoftBank Group partner with SB Energy, 2026-01-09
• Company SB Energy — SB Energy announcement (investment), 2026-01-09

xAI closed an upsized $20 billion Series E with NVIDIA and Cisco Investments as strategic investors backing its compute buildout, saying Colossus I and II ended 2025 with over one million H100 GPU equivalents. The equity and debt split was not disclosed by xAI.

With NVIDIA, Cisco Investments. Stated as “raised $20 billion”.

Related: xAI and NVIDIA (2025-10); xAI and Morgan Stanley (arranger) (2025-06-30).

• Company xAI — xAI Raises $20B Series E, 2026-01-06

Hut 8 agreed to develop at least 245 MW and up to 2,295 MW of AI data center capacity for Anthropic, starting with a 15-year, $7.0 billion lease to Fluidstack for 245 MW at River Bend, Louisiana, backstopped by Google. Anthropic said the capacity is an expansion of its work with Fluidstack and should come online by early 2027.

With Fluidstack, Google. Term: 15 years. Starts 2027. Stated as “a 15-year, $7.0 billion lease”.

Related: Anthropic and Fluidstack (2025-11-12); Fluidstack and TeraWulf (2025-08-14).

• Filing Hut 8 via SEC EDGAR — Hut 8 Signs 15-Year, 245 MW AI Data Center Lease at River Bend Campus (Exhibit 99.1), 2025-12-17
• Company Hut 8 — Hut 8 Announces AI Infrastructure Partnership with Anthropic and Fluidstack, 2025-12-17

Broadcom said on its December 2025 earnings call that Anthropic was the customer behind a $10 billion order for Google TPU Ironwood racks disclosed in September 2025, and that Anthropic placed a further $11 billion order for delivery in late 2026. Press reports in September 2025 had attributed the $10 billion order to OpenAI.

With Google. Chips: Google TPU Ironwood. Starts 2026. Stated as “a $10 billion order to sell the latest TPU ironwood racks to Anthropic; an additional $11 billion order from this same customer”.

Related: Anthropic and Google Cloud (2025-10-23); Anthropic and Google (2026-04-06); Anthropic and Broadcom (2026-10-01).

• Company The Motley Fool (via Yahoo Finance) — Broadcom (AVGO) Q4 2025 earnings call transcript, 2025-12-12
• Company The Motley Fool (via Yahoo Finance) — Broadcom (AVGO) Q4 2025 earnings call transcript (second order), 2025-12-12
• Press TrendForce — Broadcom Reportedly Secures $10B Custom AI Chip Order, 2025-09-05

HUMAIN expanded its NVIDIA partnership with plans to deploy up to 600,000 NVIDIA GPUs, including GB300 systems, in Saudi Arabia and the US over three years. The same announcement covered a 500 MW xAI data center and an AWS AI Zone in Riyadh with up to 150,000 NVIDIA GPUs.

With Global AI, xAI, Amazon Web Services. Chips: NVIDIA GB300. Term: 3 years.

Related: HUMAIN and NVIDIA (2025-05-13); xAI and HUMAIN (2025-11-19).

• Company HUMAIN (PR Newswire) — HUMAIN Expands Strategic Partnership with NVIDIA, 2025-11-19

HUMAIN and xAI agreed to jointly develop a network of data centers in Saudi Arabia anchored by a flagship facility of 500 MW or more, xAI's first large deployment outside the US.

With NVIDIA. Chips: NVIDIA GB300.

Related: HUMAIN and NVIDIA (2025-11-19); HUMAIN and NVIDIA (2025-05-13).

• Company HUMAIN (PR Newswire) — HUMAIN Expands Strategic Partnership with NVIDIA, 2025-11-19

Anthropic committed to buy $30 billion of Azure compute capacity and to contract up to 1 GW of additional capacity on NVIDIA Grace Blackwell and Vera Rubin systems. NVIDIA and Microsoft committed to invest up to $10 billion and up to $5 billion in Anthropic.

With NVIDIA. Chips: NVIDIA Grace Blackwell, NVIDIA Vera Rubin. Stated as “$30 billion of Azure compute capacity”.

Related: Anthropic and Google Cloud (2025-10-23); Anthropic and Fluidstack (2025-11-12).

• Company Anthropic — Microsoft, NVIDIA and Anthropic announce strategic partnerships, 2025-11-18
• Company Microsoft — Microsoft, NVIDIA and Anthropic announce strategic partnerships, 2025-11-18

Anthropic announced a $50 billion investment in US computing infrastructure, building custom data centers with Fluidstack in Texas and New York, with sites coming online through 2026 and more sites to come.

Starts 2026. Stated as “a $50 billion investment in American computing infrastructure”.

• 2026-04-06: Anthropic said the Google and Broadcom TPU deal, mostly sited in the US, is a major expansion of this $50 billion commitment. Source

Related: Anthropic and Google (2026-04-06); Anthropic and Hut 8 (2025-12-17); Anthropic and TeraWulf (2026-07-06); Anthropic and AMD (2026-07-22).

• Company Anthropic — Anthropic invests $50 billion in American AI infrastructure, 2025-11-12

Nebius disclosed with its third-quarter 2025 results a new agreement to deliver AI infrastructure to Meta valued at about $3 billion over five years.

Term: 5 years. Stated as “approximately $3 billion over 5 years”.

Related: Meta and Nebius (2026-03-16).

• Filing Nebius via SEC EDGAR — Nebius reports third quarter financial results (Exhibit 99.1), 2025-11-11

Cipher signed a 15-year lease with Amazon Web Services to provide 300 MW of gross capacity for AI workloads, delivered in two phases in 2026, with an estimated contract value of about $5.5 billion.

Term: 15 years. Starts 2026-08. Stated as “~$5.5Bn Estimated Contract Value”.

Related: Fluidstack and Cipher Mining (2025-09-25).

• Filing Cipher Mining via SEC EDGAR — Cipher Mining third quarter business update (Exhibit), 2025-11-03
• Filing Cipher Mining via SEC EDGAR — Cipher Mining business update (terms), 2025-11-03

IREN agreed to give Microsoft access to NVIDIA GB300 GPUs over five years for about $9.7 billion including a 20% prepayment, deployed in phases through 2026 at its Childress, Texas campus in new data centers supporting 200 MW of IT load. IREN agreed to buy the GPUs and equipment from Dell for about $5.8 billion.

With Dell Technologies. Chips: NVIDIA GB300. Term: 5 years. Stated as “a total contract value of approximately $9.7 billion, including a 20% prepayment”.

Related: Microsoft and Nebius (2025-09-08); Microsoft and Nscale (2025-10-15); Microsoft and Lambda (2025-11-03).

• Company IREN — IREN Secures $9.7bn AI Cloud Contract with Microsoft, 2025-11-03

Lambda signed a multibillion-dollar, multi-year agreement to deploy AI infrastructure for Microsoft powered by tens of thousands of NVIDIA GPUs, including GB300 NVL72 systems. The exact value and GPU count were not disclosed.

Chips: NVIDIA GB300. Stated as “multibillion-dollar agreement”.

Related: Microsoft and Nebius (2025-09-08); Microsoft and Nscale (2025-10-15); Microsoft and IREN (2025-11-03); NVIDIA and Lambda (2025-09).

• Company Lambda — Lambda Announces Multibillion-Dollar Agreement With Microsoft, 2025-11-03

AWS and OpenAI signed a $38 billion multi-year agreement giving OpenAI access to hundreds of thousands of NVIDIA GB200 and GB300 GPUs on Amazon EC2 UltraServers, with all capacity targeted for deployment before the end of 2026. The agreement was expanded by $100 billion in February 2026.

Chips: NVIDIA GB200, NVIDIA GB300. Term: 7 years. Starts 2025-11. Stated as “$38 billion agreement, which will have continued growth over the next seven years”.

Related: OpenAI and Amazon Web Services (2026-02-27).

• Company Amazon — AWS and OpenAI announce multi-year strategic partnership, 2025-11-03

OpenAI, Oracle and Related Digital announced a Stargate campus of more than a gigawatt in Saline Township, Michigan, developed exclusively for Oracle, with construction expected to begin in early 2026. The site is part of the 4.5 GW Oracle partnership.

With Related Digital. Stated as “multi-billion dollar investment”.

Related: OpenAI and Oracle (2025-07-22).

• Company Related Companies — OpenAI, Oracle and Related Digital announce Stargate data center site in Michigan, 2025-10-30

Under a restructured partnership, OpenAI contracted to purchase an incremental $250 billion of Azure services, and Microsoft gave up its right of first refusal to be OpenAI's compute provider. No term length for the commitment was stated.

Stated as “an incremental $250B of Azure services”.

Related: OpenAI and Microsoft (2023-01-23); OpenAI and Microsoft (2025-01-21).

• Company Microsoft — The next chapter of the Microsoft-OpenAI partnership, 2025-10-28

NextEra Energy agreed with Google to restart the Duane Arnold nuclear plant in Iowa, with Google buying 615 MW of its output for 25 years from the restart, expected in the first quarter of 2029.

Term: 25 years. Starts 2029.

• Company NextEra Energy — NextEra Energy and Google Announce New Collaboration to Accelerate Nuclear Energy Deployment in the U.S., 2025-10-27

Anthropic said it would expand its use of Google Cloud to up to one million TPUs, an expansion worth tens of billions of dollars that is expected to bring well over a gigawatt of capacity online in 2026.

Chips: Google TPU. Starts 2026. Stated as “worth tens of billions of dollars”.

Related: Anthropic and Google (2026-04-06); Anthropic and Broadcom (2025-12-11).

• Company Anthropic — Expanding our use of Google Cloud TPUs and Services, 2025-10-23
• Company Anthropic — Expanding our use of Google Cloud TPUs and Services (TPU count), 2025-10-23

OpenAI, Oracle and Vantage Data Centers announced a Stargate campus in Port Washington, Wisconsin, with four data centers providing close to a gigawatt of AI capacity and completion scheduled for 2028. The site is part of the 4.5 GW Oracle partnership.

With Vantage Data Centers. Stated as “Vantage's $15B+ investment in Port Washington”.

Related: OpenAI and Oracle (2025-07-22); OpenAI and Oracle (2025-09-23).

• Company Vantage Data Centers — OpenAI, Oracle and Vantage Data Centers Announce Stargate Data Center Site in Wisconsin, 2025-10-22
• Company Vantage Data Centers — Vantage announcement (investment), 2025-10-22

Meta and funds managed by Blue Owl formed a joint venture, 80% Blue Owl and 20% Meta, to develop and own the Hyperion data center campus in Richland Parish, Louisiana, with about $27 billion in total development costs, partly funded by debt sold to PIMCO and other bond investors. Meta leases the campus on four-year initial terms with extension options and a 16-year residual value guarantee.

With PIMCO. Term: 4 years. Stated as “approximately $27 billion in total development costs”.

• 2026-07-13: Meta said it is expanding the Richland Parish, Louisiana data center to 5 GW of compute capacity. Source

Related: Meta and BlackRock (2026-07-28).

• Company Meta — Meta Announces Joint Venture With Funds Managed by Blue Owl Capital to Develop Hyperion Data Center, 2025-10-21
• Company Meta — Meta-Blue Owl Hyperion joint venture (development cost), 2025-10-21

Nscale signed an expanded deal to provide Microsoft with about 200,000 NVIDIA GB300 GPUs across Texas, Portugal, the UK and Norway, including about 104,000 GPUs at a roughly 240 MW Texas campus from Q3 2026. No contract value was disclosed.

With Dell Technologies, Aker. Chips: NVIDIA GB300. Starts 2026.

Related: Microsoft and Nscale (2026-04-14); OpenAI and Nscale (2025-07-31); Microsoft and Nebius (2025-09-08); Microsoft and IREN (2025-11-03); Microsoft and Lambda (2025-11-03); Microsoft and Nscale (2026-05-05).

• Company Nscale — Nscale Contracts Approximately 200,000 NVIDIA GB300 GPUs with Microsoft, 2025-10-15

OpenAI and Broadcom signed a term sheet to deploy 10 GW of OpenAI-designed AI accelerators with Broadcom networking, in racks targeted to start deploying in the second half of 2026 and finish by the end of 2029. No dollar value was disclosed.

Chips: OpenAI-designed custom accelerator. Starts 2026.

Related: OpenAI and NVIDIA (2025-09-22); OpenAI and AMD (2025-10-06).

• Company OpenAI — OpenAI and Broadcom announce strategic collaboration to deploy 10 gigawatts of OpenAI-designed AI accelerators, 2025-10-13
• Company OpenAI — OpenAI and Broadcom announcement (term sheet), 2025-10-13

Sur Energy and OpenAI signed a letter of intent for Stargate Argentina, a data center project of up to 500 MW involving up to $25 billion of investment at full scale under Argentina's RIGI investment regime. In May 2026 the Buenos Aires Times reported the project had not yet been presented to the government.

Stated as “up to $25 billion in investment at full scale”.

• 2026-05-28: Buenos Aires Times reported that more than six months on, the project had not been presented to the government. Source

Related: OpenAI and Stargate Project (2025-01-21).

• Company Sur Energy — Stargate Argentina press release, 2025-10-10
• Company Sur Energy — Stargate Argentina press release (investment), 2025-10-10

AMD and OpenAI signed a definitive agreement for OpenAI to deploy 6 GW of AMD Instinct GPUs over multiple generations, starting with 1 GW of MI450 GPUs in the second half of 2026. AMD issued OpenAI a warrant for up to 160 million AMD shares that vests as deployment milestones are met; AMD's CFO said the partnership is expected to deliver tens of billions of dollars in revenue.

Chips: AMD MI450. Starts 2026. Stated as “warrant for up to 160 million shares of AMD common stock”.

Related: OpenAI and NVIDIA (2025-09-22); OpenAI and Broadcom (2025-10-13).

• Company AMD — AMD and OpenAI Announce Partnership to Deploy 6 Gigawatts of AMD GPUs, 2025-10-06
• Company AMD — AMD and OpenAI Announce Partnership to Deploy 6 Gigawatts of AMD GPUs (warrant terms), 2025-10-06

OpenAI signed letters of intent with Samsung and SK in Seoul under which Samsung Electronics and SK hynix will supply memory for Stargate; Samsung said OpenAI's memory demand is projected to reach up to 900,000 DRAM wafers per month. The letters also cover data center work in South Korea.

With SK hynix, Samsung SDS, Samsung C&T, Samsung Heavy Industries. Chips: DRAM (HBM) memory.

Related: OpenAI and Stargate Project (2025-01-21).

• Company Samsung Electronics — Samsung and OpenAI Announce Strategic Partnership to Accelerate Advancements in Global AI Infrastructure, 2025-10-01
• Press TechCrunch — OpenAI ropes in Samsung, SK Hynix to source memory chips for Stargate, 2025-10-01

Bloomberg reported that xAI was raising about $20 billion through a special-purpose vehicle that would buy NVIDIA chips and lease them to xAI for five years for its Colossus 2 data center, with NVIDIA investing about $2 billion of the equity. The companies did not confirm the structure.

With Apollo Global Management, Diameter Capital Partners, Valor Capital. Term: 5 years. Stated as “about $7.5 billion in equity and up to $12.5 billion in debt”.

Related: xAI and Morgan Stanley (arranger) (2025-06-30); xAI and Series E investors (2026-01-06).

• Press Quartz (citing Bloomberg), via Yahoo Finance — Elon Musk's xAI to raise $20 billion using special Nvidia-backed structure, 2025-10-08

Meta committed to pay CoreWeave up to about $14.2 billion for cloud capacity through December 14, 2031, with an option to expand through 2032. Meta exercised that option as part of a roughly $21 billion expanded agreement in April 2026.

Stated as “up to approximately $14.2 billion through December 14, 2031”.

• 2026-04-09: CoreWeave announced an expanded agreement with Meta worth about $21 billion through December 2032, including exercise of the earlier option. Source

Related: Meta and CoreWeave (2026-04-09).

• Filing CoreWeave via SEC EDGAR — CoreWeave Form 8-K (September 30, 2025), 2025-09-30

Cipher signed a 10-year agreement to host 168 MW of critical IT load for Fluidstack at Barber Lake, Texas, worth about $3 billion, with Google backstopping $1.4 billion of Fluidstack's obligations for about a 5.4% stake in Cipher. In September 2026 the lease was extended to 20 years after an unnamed leading AI lab committed to a follow-on term.

With Google. Term: 10 years. Stated as “~$3 Billion in Contracted Revenue over 10-Year Term”.

• 2025-11-20: Cipher signed an additional 56 MW, 10-year hosting agreement with Fluidstack. Source
• 2026-09-25: Barber Lake lease term extended to 20 years, raising contracted revenue to over $9 billion, with a leading AI lab committed to the added 10-year term. Source

Related: Fluidstack and TeraWulf (2025-08-14); Amazon Web Services and Cipher Mining (2025-11-03).

• Filing Cipher Mining via SEC EDGAR — Cipher Mining signs 168 MW, 10-year AI hosting agreement with Fluidstack (Exhibit 99.1), 2025-09-25
• Filing Cipher Mining via SEC EDGAR — Cipher Mining release (Google backstop), 2025-09-25

CoreWeave and OpenAI signed a new order form under which OpenAI committed to pay up to about $6.5 billion through May 31, 2031 for reserved cloud capacity. CoreWeave said this brought its total contract value with OpenAI to about $22.4 billion.

Stated as “up to approximately $6.5 billion through May 31, 2031”.

Related: OpenAI and CoreWeave (2025-03-10).

• Filing CoreWeave via SEC EDGAR — CoreWeave Form 8-K (September 25, 2025), 2025-09-25
• Company CoreWeave (Business Wire) — CoreWeave Expands Agreement with OpenAI by up to $6.5B, 2025-09-25

OpenAI, Oracle and SoftBank named five new US Stargate sites: Shackelford County, Texas, Dona Ana County, New Mexico and a Midwest site with Oracle, able to deliver over 5.5 GW together with a 600 MW Abilene expansion, plus SoftBank sites in Lordstown, Ohio and Milam County, Texas that can scale to 1.5 GW. The partners said this brought Stargate to nearly 7 GW of planned capacity and over $400 billion of investment over three years.

With SoftBank, SB Energy.

• 2025-10-22: The Midwest site was named as Port Washington, Wisconsin, developed with Vantage Data Centers. Source
• 2026-01-09: SB Energy was selected to build and operate the 1.2 GW Milam County site under a lease with OpenAI. Source

Related: OpenAI and Stargate Project (2025-01-21); OpenAI and Oracle (2025-07-22); OpenAI and Oracle (2025-10-22); OpenAI and SB Energy (2026-01-09).

• Company SoftBank Group — OpenAI, Oracle, and SoftBank expand Stargate with five new AI data center sites, 2025-09-23
• Company SoftBank Group — Five new Stargate sites (SoftBank sites), 2025-09-23

OpenAI and NVIDIA signed a letter of intent for OpenAI to deploy at least 10 GW of NVIDIA systems, with NVIDIA intending to invest up to $100 billion in OpenAI as each gigawatt is deployed. The first gigawatt was targeted for the second half of 2026 on the Vera Rubin platform.

Chips: NVIDIA Vera Rubin. Starts 2026. Stated as “NVIDIA intends to invest up to $100 billion in OpenAI progressively as each gigawatt is deployed”.

• 2026-02-25: NVIDIA's annual report said it was still finalizing an investment and partnership agreement with OpenAI, with no assurance a transaction would be completed. Source
• 2026-02-27: OpenAI said NVIDIA would invest $30 billion in its $110 billion funding round; Reuters said it was unclear whether this replaced the up to $100 billion commitment. Source
• 2026-08-17: NVIDIA agreed to guarantee up to $105 billion of lease obligations for an Ohio campus where OpenAI will deploy NVIDIA systems on about 4.25 GW of IT load. Source

Related: OpenAI and AMD (2025-10-06); OpenAI and Broadcom (2025-10-13); OpenAI and SB Energy (2026-08-17).

• Company NVIDIA — OpenAI and NVIDIA Announce Strategic Partnership to Deploy 10 Gigawatts of NVIDIA Systems, 2025-09-22

OpenAI announced Stargate UK with NVIDIA and Nscale, saying it will explore offtake of up to 8,000 GPUs in the first quarter of 2026 with potential to scale to 31,000 GPUs, across UK sites including Cobalt Park in the North East.

With NVIDIA. Starts 2026.

Related: OpenAI and Nscale (2025-07-31); Microsoft and Nscale (2025-10-15).

• Company OpenAI — Introducing Stargate UK, 2025-09-16

CoreWeave disclosed a $6.3 billion order form with NVIDIA under which NVIDIA must buy any residual cloud capacity that CoreWeave's own customers do not use, through April 13, 2032.

Stated as “an initial value of $6.3 billion”.

Related: CoreWeave and NVIDIA (2026-01-26); NVIDIA and Lambda (2025-09).

• Filing CoreWeave via SEC EDGAR — CoreWeave Form 8-K (September 15, 2025), 2025-09-15

The Wall Street Journal reported, citing people familiar with the agreement, that OpenAI signed a contract to buy about $300 billion of Oracle computing power over roughly five years starting in 2027, about seven weeks after the 4.5 GW Oracle partnership was announced. On September 23, 2025 OpenAI put that partnership at more than $300 billion over five years. Oracle has not stated the figure, and neither company has confirmed the 2027 start.

Term: 5 years. Starts 2027. Stated as “more than $300 billion over five years”.

Related: OpenAI and Oracle (2025-07-22); OpenAI and Oracle (2024-06-11).

• Company OpenAI — OpenAI, Oracle, and SoftBank expand Stargate with five new AI data center sites, 2025-09-23
• Press The Register (citing The Wall Street Journal) — OpenAI reportedly on the hook for $300B Oracle Cloud bill, 2025-09-11

Nebius agreed to provide Microsoft dedicated GPU capacity from its new Vineland, New Jersey data center over five years, for a contract value of about $17.4 billion through 2031, or about $19.4 billion if Microsoft takes additional services. Capacity was to be deployed in tranches in 2025 and 2026.

Term: 5 years. Starts 2025. Stated as “about $17.4 billion through 2031”.

Related: Microsoft and Nscale (2025-10-15); Microsoft and IREN (2025-11-03); Microsoft and Lambda (2025-11-03).

• Filing Nebius via SEC EDGAR — Nebius Group Form 6-K (September 8, 2025), 2025-09-08
• Filing Nebius via SEC EDGAR — Nebius Group Form 6-K (option), 2025-09-08

The Information reported that NVIDIA agreed to lease back about 18,000 of its own GPUs from Lambda, a company it had invested in, for about $1.5 billion: roughly $1.3 billion over four years for 10,000 GPUs plus about $200 million for 8,000 more. Neither company disclosed terms.

Term: 4 years. Stated as “$1.5 billion”.

Related: Lambda and Macquarie Group (2024-04-04); Microsoft and Lambda (2025-11-03); NVIDIA and CoreWeave (2025-09-15).

• Press Tom's Hardware (citing The Information) — Nvidia signs $1.5 billion deal with cloud startup Lambda to rent back its own AI chips, 2025-09-05

Applied Digital signed a third lease with CoreWeave for 150 MW at its Polaris Forge 1 campus in Ellendale, North Dakota, bringing CoreWeave's leases there to 400 MW and about $11 billion of anticipated lease revenue, including $7 billion from two roughly 15-year leases executed in May 2025.

Term: 15 years. Stated as “total anticipated contracted lease revenue to approximately $11B”.

Related: NVIDIA and CoreWeave (2025-09-15).

• Filing Applied Digital via SEC EDGAR — Applied Digital Finalizes Additional 150MW Lease with CoreWeave in North Dakota (Exhibit 99.1), 2025-08-29

Reuters reported, citing a source, that Google signed a six-year cloud deal with Meta worth more than $10 billion for servers, storage and networking. The deal was first reported by The Information; neither company disclosed terms.

Term: 6 years. Stated as “worth more than $10 billion”.

• Press Reuters (via CTV News) — Meta signs over US$10 billion cloud deal with Google, source says, 2025-08-21
• Press Reuters (via KSL) — Meta signs over $10 billion cloud deal with Google, source says, 2025-08-24

TeraWulf signed two 10-year colocation agreements with Fluidstack for more than 200 MW of critical IT load at Lake Mariner, New York, worth about $3.7 billion and up to $8.7 billion with extensions. Google backstopped $1.8 billion of Fluidstack's obligations and received warrants for about an 8% stake in TeraWulf.

With Google. Term: 10 years. Stated as “approximately $3.7 billion in contracted revenue”.

• 2025-08-18: TeraWulf announced a Fluidstack expansion with a 160 MW CB-5 lease at Lake Mariner. Source

Related: Fluidstack and Cipher Mining (2025-09-25); Anthropic and Hut 8 (2025-12-17).

• Filing TeraWulf via SEC EDGAR — TeraWulf Signs 200+ MW, 10-Year AI Hosting Agreements (Exhibit 99.1), 2025-08-14
• Filing TeraWulf via SEC EDGAR — TeraWulf release (Google backstop), 2025-08-14

CoreWeave closed a $2.6 billion secured delayed draw term loan (DDTL 3.0) led by Morgan Stanley and MUFG to buy equipment and infrastructure for its long-term contract with OpenAI.

With Goldman Sachs, OpenAI. Stated as “$2.6 billion delayed draw term loan facility”.

Related: OpenAI and CoreWeave (2025-03-10).

• Filing CoreWeave via SEC EDGAR — CoreWeave Closes $2.6 Billion Secured Debt Financing Facility (Exhibit 99.1), 2025-07-31
• Filing CoreWeave via SEC EDGAR — CoreWeave DDTL 3.0 release (use of funds), 2025-07-31

Nscale, Aker and OpenAI announced Stargate Norway in Narvik, planned at 230 MW with an ambition for 290 MW more and a target of 100,000 NVIDIA GPUs by the end of 2026, with OpenAI as an initial offtaker. Nscale and Aker committed about $1 billion to the first 20 MW phase; the Aker-Nscale venture later contracted Narvik capacity to Microsoft.

With Aker. Stated as “approximately USD 1 billion to the initial 20MW phase of the project”.

• 2025-10-15: Nscale said the Aker-Nscale joint venture had signed a multi-year agreement to deliver about 52,000 NVIDIA GB300 GPUs to Microsoft from Narvik. Source

Related: Microsoft and Nscale (2025-10-15); Microsoft and Nscale (2026-04-14); OpenAI and Stargate Project (2025-01-21).

• Company Nscale — Nscale, Aker and OpenAI to establish Stargate Norway, 2025-07-31

Oracle and OpenAI agreed to develop 4.5 GW of additional Stargate data center capacity in the US, which OpenAI said would bring Stargate capacity under development to over 5 GW including Abilene. On June 30, 2025 Oracle had disclosed an unnamed cloud contract expected to contribute more than $30 billion in annual revenue starting in fiscal 2028.

• 2025-09-23: OpenAI put the 4.5 GW partnership at more than $300 billion between the two companies over five years. Source
• 2025-10-22: Port Washington, Wisconsin campus with Vantage named as part of the 4.5 GW partnership. Source
• 2025-10-30: Saline Township, Michigan campus with Related Digital named as part of the 4.5 GW partnership. Source

Related: OpenAI and Oracle (2025-09-10); OpenAI and Oracle (2024-06-11); OpenAI and Stargate Project (2025-01-21); OpenAI and Oracle (2025-09-23); OpenAI and Oracle (2025-10-22); OpenAI and Oracle (2025-10-30); Crusoe and Blue Owl Capital (2025-05-21).

• Company OpenAI — Stargate advances with 4.5 GW partnership with Oracle, 2025-07-22
• Filing Oracle via SEC EDGAR — Oracle Form 8-K (June 30, 2025), 2025-06-30

Morgan Stanley said xAI raised $5 billion of secured notes and term loans plus $5 billion of strategic equity, with proceeds going to AI infrastructure including one of the world's largest data centers. The figures come from the arranging bank as reported by press, not from xAI.

Stated as “$10 billion in debt and equity”.

Related: xAI and NVIDIA (2025-10); xAI and Series E investors (2026-01-06).

• Press TechCrunch — xAI raises $10B in debt and equity, 2025-07-01

Talen Energy expanded its nuclear power agreement with Amazon to up to 1,920 MW from the Susquehanna plant in Pennsylvania through 2042, reaching the full volume no later than 2032.

• Filing Talen Energy (SEC filing) — Talen Energy Expands Nuclear Energy Relationship with Amazon (Exhibit 99.1), 2025-06-11

Reuters reported, citing three sources, that OpenAI would add Google Cloud capacity for training and running its models in a deal finalized in May 2025. Neither company disclosed terms.

Related: OpenAI and Google Cloud (2025-06).

• Press Reuters (via South China Morning Post) — OpenAI taps Google in unprecedented cloud deal despite AI rivalry, sources say, 2025-06-11

Meta signed a 20-year agreement for the output of Constellation's Clinton nuclear plant in Illinois, 1,121 MW including a 30 MW uprate, starting in June 2027.

Term: 20 years. Starts 2027-06.

• Company Constellation Energy — Constellation, Meta Sign 20-Year Deal for Clean, Reliable Nuclear Energy in Illinois, 2025-06-03

Reuters and other outlets reported in late June 2025 that OpenAI had begun renting Google TPUs. OpenAI then said it was only in early testing with some TPUs and had no active plans to deploy them at scale.

Chips: Google TPU.

Related: OpenAI and Google Cloud (2025-06-10).

• Press Reuters (via China Daily Hong Kong) — OpenAI says it has no plans to use Google's in-house chip, 2025-07-01

OpenAI announced Stargate UAE with G42, Oracle, NVIDIA, Cisco and SoftBank: a 1 GW cluster in Abu Dhabi, with the first 200 MW expected to go live in 2026. In October 2025 G42 said construction of the first 200 MW was well underway.

With Oracle, NVIDIA, Cisco, SoftBank. Starts 2026.

• 2025-10-16: G42 said construction of the first 200 MW was well underway toward planned 2026 delivery. Source

Related: OpenAI and Stargate Project (2025-01-21).

• Company OpenAI — Introducing Stargate UAE, 2025-05-22
• Company G42 (PR Newswire) — G42 provides update on construction of Stargate UAE AI infrastructure cluster, 2025-10-16

Crusoe, Blue Owl and Primary Digital Infrastructure began the second phase of a $15 billion joint venture to fund the 1.2 GW Abilene, Texas campus, adding six buildings for eight in total. The campus runs on Oracle Cloud Infrastructure as Stargate's flagship site for OpenAI, with the first NVIDIA GB200 racks delivered in June 2025.

With Primary Digital Infrastructure, Oracle, OpenAI. Chips: NVIDIA GB200. Stated as “a $15 billion joint-venture”.

• 2025-09-30: Crusoe said the first phase of the Abilene campus was live on Oracle Cloud Infrastructure. Source
• 2026-09-10: Oracle said 6 of the 8 Abilene buildings, 618 MW and 75% of capacity, had been delivered to the customer. Source

Related: OpenAI and Stargate Project (2025-01-21); OpenAI and Oracle (2025-07-22).

• Company Crusoe — Crusoe, Blue Owl Capital, and Primary Digital Infrastructure enter second phase of $15 billion joint venture, 2025-05-21
• Company Crusoe — Crusoe announces flagship Abilene data center is live, 2025-09-30

Saudi Arabia's HUMAIN and NVIDIA announced plans for AI factories of up to 500 MW powered by several hundred thousand NVIDIA GPUs over five years, starting with an 18,000-GPU GB300 supercomputer. The partnership was expanded in November 2025.

Chips: NVIDIA GB300. Term: 5 years.

Related: HUMAIN and NVIDIA (2025-11-19); xAI and HUMAIN (2025-11-19).

• Company NVIDIA — HUMAIN and NVIDIA Announce Strategic Partnership to Build AI Factories of the Future in Saudi Arabia, 2025-05-13

AMD and Saudi Arabia's HUMAIN announced a $10 billion collaboration to deploy 500 MW of AI compute built on AMD hardware.

Stated as “a $10 billion collaboration”.

• Company AMD — AMD and HUMAIN Form Strategic, $10B Collaboration to Advance Global AI, 2025-05-13

CoreWeave agreed to supply AI cloud capacity to OpenAI under a contract worth up to $11.9 billion through October 2030, and OpenAI received $350 million of CoreWeave stock as part of the deal. CoreWeave later financed equipment for the contract with a $2.6 billion delayed draw term loan.

Stated as “up to $11.9 billion”.

• 2025-05: CoreWeave later said the contract was expanded by up to $4 billion in May 2025. Source
• 2025-09-25: OpenAI committed up to about $6.5 billion more through May 31, 2031 under a new order form. Source

Related: OpenAI and CoreWeave (2025-09-25); CoreWeave and Morgan Stanley and MUFG (lead arrangers) (2025-07-31).

• Company CoreWeave — CoreWeave Announces Agreement with OpenAI to Deliver AI Infrastructure, 2025-03-10
• Filing CoreWeave via SEC EDGAR — CoreWeave IPO prospectus (Form 424B4), 2025-03-31

Microsoft and OpenAI replaced Azure exclusivity on new OpenAI capacity with a Microsoft right of first refusal, and Microsoft approved OpenAI building additional capacity, primarily for research and training. The change was announced alongside the Stargate Project.

• 2025-10-28: The right of first refusal on OpenAI compute ended under the restructured agreement. Source

Related: OpenAI and Microsoft (2023-01-23); OpenAI and Microsoft (2025-10-28); OpenAI and Stargate Project (2025-01-21).

• Company Microsoft — Microsoft and OpenAI evolve partnership to drive the next phase of AI, 2025-01-21

OpenAI, SoftBank, Oracle and MGX launched the Stargate Project, a new company that intends to invest $500 billion over four years in AI infrastructure for OpenAI in the United States, starting with $100 billion. SoftBank has financial responsibility and OpenAI operational responsibility.

With SoftBank, Oracle, MGX, Arm, Microsoft, NVIDIA. Term: 4 years. Starts 2025. Stated as “$500 billion over the next four years”.

• 2025-07-22: Oracle partnership for 4.5 GW of additional US capacity brought Stargate capacity under development to over 5 GW. Source
• 2025-09-23: Five new sites brought Stargate to nearly 7 GW of planned capacity and over $400 billion of investment over three years. Source

Related: OpenAI and Microsoft (2025-01-21); OpenAI and Oracle (2025-07-22); OpenAI and Oracle (2025-09-23); Crusoe and Blue Owl Capital (2025-05-21).

• Company OpenAI — Announcing The Stargate Project, 2025-01-21

AWS announced Project Rainier, an EC2 UltraCluster of Trainium2 UltraServers built with Anthropic and spanning hundreds of thousands of Trainium2 chips. AWS said in October 2025 that it was fully operational with nearly half a million Trainium2 chips.

Chips: AWS Trainium2.

• 2025-10-29: AWS said Project Rainier was fully operational and that Claude would run on more than 1 million Trainium2 chips by the end of 2025. Source

Related: Anthropic and Amazon Web Services (2024-11-22); Anthropic and Amazon Web Services (2026-04-20).

• Company Amazon — AWS Trainium2 instances now generally available, 2024-12-03
• Company Amazon — AWS activates Project Rainier, 2025-10-29

Amazon invested a further $4 billion in Anthropic, bringing its total to $8 billion, and AWS became Anthropic's primary cloud and training partner, with Anthropic working with Annapurna Labs on future Trainium chips.

With Amazon. Chips: AWS Trainium. Stated as “a new $4 billion investment from Amazon”.

Related: Anthropic and Amazon Web Services (2023-09-25); Anthropic and Amazon Web Services (2024-12-03); Anthropic and Amazon Web Services (2026-04-20).

• Company Anthropic — Powering the next generation of AI development with AWS, 2024-11-22

NVIDIA said xAI's Colossus cluster in Memphis, Tennessee comprised 100,000 NVIDIA Hopper GPUs on Spectrum-X Ethernet, built in 122 days, and that xAI was doubling it to 200,000 Hopper GPUs. Purchase terms were not disclosed; in May 2026 Anthropic contracted all of Colossus 1's capacity.

Chips: NVIDIA H100. Starts 2024.

• 2026-05-06: Anthropic signed an agreement to use all of the compute capacity at Colossus 1, more than 300 MW. Source

Related: Anthropic and SpaceX (2026-05-06); xAI and Morgan Stanley (arranger) (2025-06-30); xAI and NVIDIA (2025-10).

• Company NVIDIA — NVIDIA Spectrum-X Makes Colossal NVIDIA Hopper 100,000-GPU System Possible, 2024-10-28

Google agreed to buy power from several small modular reactors to be developed by Kairos Power, up to 500 MW in total, with the first reactor targeted for 2030 and the rest through 2035.

• Company Google — New nuclear clean energy agreement with Kairos Power, 2024-10-14

Constellation signed a 20-year power purchase agreement with Microsoft to restart Three Mile Island Unit 1 as the Crane Clean Energy Center, adding about 835 MW of carbon-free power for Microsoft's PJM data centers, expected online in 2028.

Term: 20 years. Starts 2028.

• Company Constellation Energy — Constellation to Launch Crane Clean Energy Center, 2024-09-20

Oracle, Microsoft and OpenAI agreed to extend the Microsoft Azure AI platform to Oracle Cloud Infrastructure to give OpenAI additional capacity. No contract value or capacity figure was disclosed.

With Microsoft.

Related: OpenAI and Oracle (2025-07-22); OpenAI and Oracle (2025-09-10).

• Company Oracle — OpenAI Selects Oracle Cloud Infrastructure to Extend Microsoft Azure AI Platform, 2024-06-11

CoreWeave signed a $7.5 billion debt facility led by Blackstone, with Magnetar and Coatue, to expand its GPU fleet for existing customer contracts; CoreWeave called it one of the largest private financings in history.

With Magnetar, Coatue, Carlyle, CDPQ, DigitalBridge Credit, BlackRock, Eldridge Industries, Great Elm Capital. Stated as “a $7.5 billion debt financing facility”.

Related: CoreWeave and Magnetar Capital and Blackstone (2023-08-03); CoreWeave and Morgan Stanley and MUFG (lead arrangers) (2025-07-31).

• Company CoreWeave — CoreWeave Secures $7.5 Billion Debt Financing Facility led by Blackstone and Magnetar, 2024-05-17

Lambda secured a GPU-backed special purpose financing vehicle of up to $500 million, led by Macquarie, to expand its on-demand NVIDIA GPU cloud.

With Industrial Development Funding. Stated as “a special purpose GPU financing vehicle of up to $500 million”.

Related: NVIDIA and Lambda (2025-09); Microsoft and Lambda (2025-11-03).

• Company Lambda (Business Wire) — Lambda Announces $500M GPU-Backed Facility to Expand Cloud for AI, 2024-04-04

Amazon agreed to invest up to $4 billion in Anthropic, and AWS became Anthropic's primary cloud provider for mission-critical workloads, with access to AWS Trainium and Inferentia chips.

With Amazon. Chips: AWS Trainium, AWS Inferentia. Stated as “Amazon will invest up to $4 billion in Anthropic”.

Related: Anthropic and Amazon Web Services (2024-11-22); Anthropic and Amazon Web Services (2024-12-03); Anthropic and Amazon Web Services (2026-04-20).

• Company Anthropic — Expanding access to safer AI with Amazon, 2023-09-25

CoreWeave secured a $2.3 billion debt facility led by Magnetar and Blackstone, committed to buying hardware for customer contracts it had already signed. Coatue, DigitalBridge, BlackRock, PIMCO, Carlyle and Great Elm also participated.

With Coatue, DigitalBridge, BlackRock, PIMCO, Carlyle, Great Elm. Stated as “a $2.3 billion debt facility”.

Related: CoreWeave and Blackstone (2024-05-17).

• Company CoreWeave — CoreWeave Secures $2.3 Billion Debt Financing Facility led by Magnetar Capital and Blackstone, 2023-08-03

CNBC reported that Microsoft agreed to spend potentially billions of dollars over multiple years on GPU cloud capacity from Nvidia-backed CoreWeave to help meet demand from OpenAI. Neither company disclosed terms at the time.

With OpenAI. Stated as “could be worth billions of dollars over multiple years”.

Related: OpenAI and Microsoft (2023-01-23).

• Press CNBC — Microsoft inks deal with CoreWeave to meet OpenAI cloud demand, 2023-06-01

Microsoft announced the third phase of its OpenAI partnership, a multiyear, multibillion-dollar investment under which Azure became OpenAI's exclusive cloud provider for research, products and API services. The cloud terms were changed in January 2025 and again in October 2025.

Stated as “multiyear, multibillion dollar investment”.

• 2025-01-21: Exclusivity on new capacity replaced by a Microsoft right of first refusal. Source
• 2025-10-28: OpenAI contracted an incremental $250B of Azure services and the right of first refusal ended. Source

Related: OpenAI and Microsoft (2025-01-21); OpenAI and Microsoft (2025-10-28).

• Company Microsoft — Microsoft and OpenAI extend partnership, 2023-01-23