OpenAI signs 457MW TensorWave deal as AMD buildout faces delivery test
TensorWave will supply dedicated AMD infrastructure under a six-year OpenAI agreement, with delivery expected from 2027. Its operating record and financial backing sit alongside an undisclosed contract price and a recently terminated site lease.
- TensorWave’s six-year OpenAI agreement covers 457 megawatts of dedicated AMD infrastructure across multiple U.S. sites.
- Delivery is expected to start in the first half of 2027 and finish by early 2028; contract pricing and sites remain undisclosed.
- The deal advances an existing AMD partnership, while a separate TensorWave site lease has recently ended.
OpenAI, the AI model developer, has signed a six-year agreement for 457 megawatts of dedicated infrastructure from TensorWave, a Las Vegas cloud provider focused on AMD graphics processors. TensorWave announced its largest deployment to date on November 16, putting an operator behind part of OpenAI’s planned expansion with chipmaker AMD.
The agreement comes with delivery still ahead. TensorWave has experience running successive generations of AMD hardware and financial backing from AMD’s venture arm. But its wider infrastructure pipeline also includes a recently terminated lease, making the distinction between secured capacity and operating compute consequential.

Company-reported: TensorWave’s OpenAI agreement covers 457 megawatts over six years. Graphic: Magica; source: TensorWave. Source: TensorWave.
An operator for an existing AMD expansion
TensorWave says in its announcement that it is the first publicly announced neocloud infrastructure provider supporting OpenAI’s strategic partnership with AMD. Neoclouds specialize in GPU computing for AI; the designation applies to this collaboration.
TensorWave will deploy AMD Instinct MI455X GPUs in Helios rack-scale systems, which group processors and supporting components into an integrated computing platform. High-performance Ethernet will connect the racks for large-scale model training and inference, the process of running models to generate outputs. The infrastructure will serve OpenAI alone.
The relationship predates this contract. OpenAI and AMD announced a six-gigawatt agreement on October 6, 2025, spanning multiple GPU generations. Their hardware and software collaboration began with MI300X and continued with MI350X. That announcement scheduled the first one-gigawatt MI450 deployment to begin in the second half of 2026; it does not establish how much is operating today.
