Tuesday, Nov 17 | --:--
Back to home

TensorWave Says It Will Supply OpenAI 457 MW of AMD GPU Capacity Under a Six-Year Deal

TensorWave, an AMD-backed AI cloud, announced on November 16 a six-year agreement to provide OpenAI with 457 megawatts of dedicated IT capacity across multiple US data centre sites, powered by AMD Instinct MI455X GPUs in Helios rack-scale systems. Capacity is expected to begin coming online in the first half of 2027, with full deployment by early 2028. TensorWave calls itself the first neocloud publicly announced as a provider under OpenAI's partnership with AMD; the release does not disclose contract value, site locations or financing, and OpenAI is quoted only through a compute executive.

4 min read Las Vegas, Nevada View as Markdown
A vast sage stone hearth is neatly stacked with unlit cut-paper logs, a single unstruck match with a dark red head resting on the hearth edge, with an official TensorWave logo paper clipping glued along the front of the hearth slab and an official OpenAI logo paper clipping tucked against the lintel above the right side of the hearth.

OpenAI's AMD build-out now has a named landlord, but not yet a price or an address. TensorWave, a cloud that runs only on AMD GPUs, says it will provide OpenAI with 457 megawatts of dedicated capacity under a six-year agreement, none of which is expected to be online before the first half of 2027. For AMD it is evidence that its Helios racks can be sold through neocloud operators, not only direct. For readers, it is a capacity commitment on a calendar, not compute that exists today.

What TensorWave announced#

In a November 16 release, TensorWave said it will supply 457 megawatts of IT capacity across multiple US data centre sites, powered by AMD Instinct MI455X GPUs. The infrastructure will be built for AMD's Helios rack-scale systems and run as dedicated, single-tenant capacity for OpenAI, with Ethernet scale-out networking between racks. Capacity is expected to begin coming online in the first half of 2027, with the full deployment expected by early 2028.

TensorWave says the agreement is part of OpenAI and AMD's previously announced strategic partnership and that it is the first neocloud publicly announced as an infrastructure provider supporting it. It calls the deal its largest deployment to date, and says it raised a Series B in June 2026 and has secured more than 2 gigawatts of long-term data centre capacity.

Both partners are quoted. OpenAI's chief technology officer for compute, Uday Ruddarraju, said the company is "excited to advance our previously announced partnership with AMD through this agreement with TensorWave, which will help us continue scaling AI to serve more people and organizations around the world." AMD's senior vice president for AI, Vamsi Bopanna, called it a "landmark deployment".

What is still missing#

The release gives no contract value, no site locations, and no account of how TensorWave will finance the build. It does not say how the 457 MW relates to the capacity OpenAI and AMD have previously discussed. Helios is a young product; our AMD third-quarter report covers where AMD said it stood on the rack at its last earnings, and the release does not address whether supply can scale to meet a 2027 start.

Limits#

  • Everything here comes from TensorWave's own release, which includes statements from OpenAI and AMD executives. Reuters' version on TradingView carries the same release text, so it is not an independent check.
  • "First neocloud" and "largest deployment" are TensorWave's characterisations. We have not seen OpenAI or AMD confirm them separately.
  • The timeline is expected, not contracted in anything we read; we have not seen the agreement.

Sources#

Prior Coverage

Earlier Times of AI reporting on this thread.

Scroll to continue reading