OpenAI is joining forces with TensorWave, a U.S. artificial intelligence (AI) cloud company, to secure AMD-based AI infrastructure on a large scale. The move is meant to diversify a graphics processing unit (GPU) supply chain centered on Nvidia and to respond to surging AI training and inference demand. It is drawing attention because the strategic partnership with AMD signed last year is now leading to actual infrastructure buildout.
TensorWave contract
TensorWave announced on its official website on November 16 (local time) that it signed a six-year contract to supply OpenAI with a total of 457 megawatts (MW) of dedicated AI infrastructure. It plans to deploy AMD's next-generation GPU, the 'Instinct MI455X', at several data centers in the United States and build infrastructure based on the rack-scale system 'Helios'. The first operation is set for the first half of 2027, and the full buildout is scheduled to be completed by early 2028. The contract value was not disclosed.
The contract was pursued on the basis of the GPU supply agreement of 6 gigawatts (GW) in total that OpenAI signed with AMD in October last year. At the time, the two companies agreed to adopt AMD GPUs across several generations and announced a plan to begin deploying the first 1 GW of infrastructure from the second half of this year. TensorWave is the first company officially announced among the neocloud providers taking part in that cooperation.
What it means for OpenAI and AMD
For OpenAI, the deal makes it possible to spread out a GPU procurement structure concentrated on Nvidia and to secure additional large-scale computing resources. Once the AMD-based infrastructure is in full operation, OpenAI is expected to lower its dependence on a particular chip supplier and widen its options for GPU procurement and infrastructure operation.
The contract also appears to be an important business expansion opportunity for AMD. AMD, which has been regarded as an alternative to Nvidia, can now secure commercial infrastructure that handles real large-scale AI workloads. In particular, with the MI455X and Helios, it has widened its base to compete with Nvidia in the market for rack-level AI infrastructure, beyond individual GPUs.
Neocloud providers and funding
The role of neocloud providers is also growing. TensorWave is an AI cloud company specialized in AMD GPUs, and its main business is building and operating large GPU clusters. This contract is the largest since the company's founding, and it reflects a trend of AI developers securing computing resources through specialized cloud providers as well as their own data centers.
However, large-scale infrastructure expansion requires substantial funding and a stable power supply. TensorWave said it has secured more than 2 GW of long-term data center capacity in the United States, but it did not disclose the specific investment needed for this contract or how it will raise the funds. Because AMD's venture investment arm is backing TensorWave, the capital and business ties among AI chip companies, cloud providers and AI developers are also becoming closer, which is worth noting.
Piotr Tomasik, president and co-founder of TensorWave, said, "As AI infrastructure enters a new phase, the ability to supply hundreds of MW of advanced computing resources reliably and quickly is becoming important," and added, "Through this project, we will supply a significant part of the core infrastructure that supports next-generation AI computing for OpenAI and AMD."
Original Korean article: ZDNet Korea
English brief by Korea Wind Intelligence, koreawindintel.com















