OpenAI has signed a brand new take care of AMD because it appears to diversify its sourcing of next-generation AI chips.
The deal will see OpenAI take 6GW price of AMD’s Intuition GPUs, which is able to span a number of generations and assist energy its future AI infrastructure. It plans to take the primary 1GW in H2 2026, with the chips all set to be constructed on AMD’s Intuition MI450.
Underneath the settlement between the 2 firms, AMD turns into a core strategic compute associate, supplying rack-scale AI programs and future GPU generations by means of a co-optimised {hardware} and software program roadmap. The businesses say the collaboration deepens an current relationship that began with MI300X and continued with the MI350X collection.
Moreover, as a part of the deal, AMD has issued OpenAI a warrant for as much as 160 million AMD frequent shares. The warrant vests in tranches towards particular milestones, together with the preliminary 1GW deployment and subsequent purchases as much as 6GW, and can be tied to AMD share-price targets and OpenAI’s technical and industrial milestones required to scale AMD deployments. The train worth is $0.01. That might give OpenAI a ten% stake within the chipmaker.
“We’re thrilled to associate with OpenAI to ship AI compute at huge scale,” stated Dr. Lisa Su, chair and CEO, AMD.
“This partnership brings one of the best of AMD and OpenAI collectively to create a real win-win enabling the world’s most bold AI buildout and advancing your entire AI ecosystem.”
“This partnership is a significant step in constructing the compute capability wanted to understand AI’s full potential,” added Sam Altman, co-founder and CEO of OpenAI.
“AMD’s management in high-performance chips will allow us to speed up progress and produce the advantages of superior AI to everybody quicker.”
OpenAI courts each chip giants
The settlement between OpenAI and AMD comes sizzling on the heels of the AI large’s strategic partnership with Nvidia, which it introduced final month. That can see OpenAI deploy a minimum of 10GW of Nvidia programs for its next-generation AI infrastructure, with the preliminary part of that deployment set to return on-line within the second half of 2026 on Nvidia’s Vera Rubin platform.
That deal additionally comes with a hefty funding from Nvidia, which is committing as much as $100 billion to help the development of OpenAI’s next-generation AI knowledge centres. The corporate can be set to obtain non-voting shares within the maker of ChatGPT as a part of the deal.
It’s unsurprising to see OpenAI purchase chips from each AMD and Nvidia, in any case the corporate is trying to develop its compute capability exponentially. With that explosive progress clearly comes constraints on when it’s going to really obtain chips, particularly as firms comparable to Google, Meta and Microsoft may even all be clamouring to get their arms on the most recent and best AI chips.
In truth, the offers signed with AMD and Nvidia don’t even get OpenAI to the capability it’s focusing on – which is estimated to be round 23GW of recent compute. That’s why the corporate has additionally signed a take care of Oracle to lease 4.5 GW of computing energy at a price of $30 billion.
All of those agreements imply one factor, nevertheless – a really massive invoice for OpenAI. In truth, the corporate estimates that to get to its focused capability it’s going to doubtless must spend over $1 trillion. That’s numerous money for a corporation that’s nonetheless making a loss and attaining annual revenues of simply $13 billion.
