TensorWave will supply OpenAI with 457MW of AMD-powered AI capacity

The TensorWave OpenAI deal runs for six years on AMD’s MI455X chips and Helios racks, with first capacity due in 2027. Neither side disclosed its value.


TensorWave will supply OpenAI with 457MW of AMD-powered AI capacity
Image Credits Credit: TensorWave

TensorWave, an AI cloud provider backed by AMD’s venture arm, will supply OpenAI with 457 megawatts of computing capacity. The deal runs for six years. The capacity will run on AMD’s Instinct MI455X GPUs across several US data centre sites, TensorWave said on Monday.

The TensorWave OpenAI deal falls under OpenAI’s existing partnership with AMD. A neocloud is a cloud company built around GPUs for AI. TensorWave is the first one that the partners have publicly named as a provider under that deal. Neither company gave a value for the contract or named the sites.

The first capacity should come online in the first half of 2027. TensorWave expects to finish the full deployment by early 2028.

Built for Helios

TensorWave is designing the sites for AMD’s Helios racks, which group GPUs into a single rack-scale system. Ethernet will link the racks, and the capacity will serve OpenAI alone, with no other tenants.

“We’re excited to advance our previously announced partnership with AMD through this agreement with TensorWave,” said Uday Ruddarraju, OpenAI’s chief technology officer for compute.

OpenAI put Ruddarraju in charge of its data centre and compute work in October.

TensorWave’s biggest deployment

The Las Vegas company runs only AMD hardware. It was the first public cloud to deploy AMD’s MI300X GPUs. It has since added the MI325X and MI355X, its blog post said. The OpenAI deal is its largest deployment so far.

TensorWave raised its $350m Series B in June. Its investors include Magnetar, AMD Ventures and Nexus Venture Partners. The company said it has secured more than 2 gigawatts of long-term data centre capacity.

“AI infrastructure is entering a new phase, where the ability to deliver hundreds of megawatts of advanced compute reliably and at speed will separate the companies that can support frontier AI from those that cannot,” said Piotr Tomasik, TensorWave’s president and co-founder.

Vamsi Boppana, AMD’s senior vice president for AI, called it a “landmark deployment” for OpenAI in the same announcement.

Get the TNW newsletter

Get the most important tech news in your inbox each week.

Thanks! Check your inbox to confirm your subscription.

By subscribing you agree to our Privacy Statement.

Published
Back to top