Partnership
TensorWave x OpenAI (457 MW AMD capacity agreement)
TensorWave signed a six-year agreement to provide OpenAI with 457 megawatts of dedicated, single-tenant IT capacity across multiple U.S. data center sites, running on AMD Instinct MI455X GPUs in AMD Helios racks, under OpenAI's existing partnership with AMD. Capacity is expected to start coming online in the first half of 2027, with full deployment by early 2028. Sites and contract value were not disclosed.
deployment · Announced · just announced
Machine-readable surfaces
- Markdown mirror: /partnerships/tensorwave-x-openai-457mw-amd-capacity-2026.md
- JSON-LD: embedded in this page’s head
- REST API: /v1/partnerships/tensorwave-x-openai-457mw-amd-capacity-2026
- Revision history: /partnerships/tensorwave-x-openai-457mw-amd-capacity-2026/history
- Data documentation: /data
- Query this programmatically: Deploy MCP
d5b77099-333d-4e60-b32d-2a8306cbfbab- Type
- deployment
- Status
- Announcedjust announced
- Announced
- 2026-11-163 days ago
- Active for
- 3 days
Who’s involved
TensorWave is an AMD-exclusive GPU cloud built as a deliberate counterweight to the Nvidia-dominated neocloud pack. Founded in 2023 and headquartered in Las Vegas, the company was among the first commercial clouds to stand up AMD Instinct MI300X capacity at scale, and it has since expanded its fleet to include MI325X, MI355X, and MI455X accelerators. Its pitch to model labs and inference-heavy enterprises is straightforward: MI300-class parts carry more HBM per socket than an H100 or H200, which lets a single node hold larger models resident in memory, and TensorWave is one of the few operators with the AMD supply relationship, ROCm engineering staff, and datacenter footprint to actually rent that capacity by the hour. The company positions itself squarely as an alternative to what CEO Darrick Horton has publicly called the "GPU monopoly." It signed a 1 GW infrastructure agreement with Tecfusions in 2024, opened a Las Vegas headquarters campus in April 2026, and in August 2026 disclosed a $6.5 billion contract with Fermi America, one of the largest single commitments any AMD-first cloud has landed. Named partners and tenants include Meta, AMD itself, Moreh, Modular, Luma AI, Featherless AI, MangoBoost, Wafer AI, and Edgecore. TensorWave has raised more than $500 million across a seed, a $43 million round in October 2024, a $100 million round in 2025, and a subsequent raise that valued the company at roughly $1.55 billion in June 2026. It remains private. The founding team is drawn from datacenter, semiconductor, and cloud infrastructure backgrounds; Horton was named to the Forbes 30 Under 30 AI list for 2026.
Most-frequent partners
OpenAI is a San Francisco based AI research and product company that has become one of the largest committed buyers and co-developers of datacenter capacity in the world. Founded in December 2015 by Sam Altman, Greg Brockman, Elon Musk, Ilya Sutskever and others, it operates as OpenAI Group PBC (a public benefit corporation) after a 2025 restructuring, with Microsoft holding roughly a 27 percent stake and continuing as its principal cloud partner. In June 2026 the company filed for an IPO with the SEC. On the infrastructure side, OpenAI is a 40 percent co-owner of Stargate LLC, the joint venture announced with SoftBank, Oracle and Abu Dhabi's MGX in January 2025 to build AI training and inference campuses. Stargate targets 10 GW of capacity, with initial site build-out concentrated in Abilene, Texas and five additional US sites announced in September 2025 (Shackelford County TX, Milam County TX, Doña Ana County NM, Lordstown OH, plus one undisclosed Midwest site). International Stargate sites are under exploration in the UK, Norway, Japan, the UAE and Argentina (up to 500 MW in Patagonia). The Stargate arrangement is unusual in that the capacity is contracted exclusively to OpenAI as tenant: OpenAI is simultaneously a co-owner of the operating vehicle and its sole workload. Silicon commitments span Nvidia (initial partner), AMD (up to 6 GW of Instinct GPUs) and Broadcom (10 GW of custom accelerators). This makes OpenAI, alongside Microsoft, one of the two anchor demand-side counterparties reshaping US grid siting decisions in 2025 and 2026.
Related partnerships
Other partnerships of OpenAI
- investment2025-12-03operating
- Vantage Data Centers x OpenAI x Oracle (Stargate Wisconsin)Vantage Data Centers (OpenAI / Oracle) + OpenAI + Oracledeployment2025-10-22just announced
- Stargate UAE (G42 x OpenAI x Oracle x NVIDIA x Cisco x SoftBank)OpenAI + Oracle + SoftBank + NVIDIA + G42 + Ciscojoint venture2025-05-22just announced
- joint venture2025-01-21operating
- investment2024-10-02operating
Recent partnerships of the same kind
- Active Ants x AutoStoreAutoStore + Active Antsdeployment—operating
- deployment—just announced
- deployment—operating
- deployment—operating
- deployment—operating
- aethon x Mercy HospitalAethon + Mercy Hospitaldeployment—operating
Sources (2)
Deploy Watch
Track this partnership.
We notify you when the status changes against primary-source evidence, expansion is verified, or the partnership is terminated.