Sam Altman, CEO of OpenAI, Lisa Su, CEO of Advanced Micro Devices, and Michael Intrator, CEO of CoreWeave,
CQ-Roll Call, Inc via Getty Images
AMD just won the biggest AI deal in the company’s history, with OpenAI entering a definitive agreement to deploy 6 gigawatts of AMD GPUs, with the first gigawatt of capacity powered by AMD Instinct MI450 series GPUs set to start in the second half of 2026.
The multi-generation commitment positions AMD to drive large-scale deployments to power OpenAI’s next-generation AI infrastructure alongside the massive and ongoing OpenAI investments in Nvidia hardware. The announcement is expected to generate tens of billions of dollars in AI revenue annually for AMD over time while accelerating OpenAI’s infrastructure buildout.
What Did AMD OpenAI announce?
The agreement goes beyond a massive purchase, and could lead to OpenAI having up to 10% ownership in AMD common stock. Key points of the agreement include:
- A multi-year, multi-generation partnership enabling 6GW of AMD Instinct GPU deployments to power OpenAI’s next-gen AI infrastructure.
- AMD has issued OpenAI a performance-based warrant for up to 160 million shares of AMD stock, structured to vest as specific milestones are achieved. This move will help ensure strategic and product alignment between the two companies going forward.
- First deployments will begin 2H 2026, anchored on AMD’s next-generation Instinct MI450 GPUs and rack-scale systems.
What’s the Big Deal?
The deal accomplishes two things critical to AMD’s AI growth strategy in addition to significant revenue. First, it adds tremendous credibility to AMD’s GPUs, which can deliver competitive performance. But the deal will also help AMD improve its AI software story, as OpenAI ports and enhances its AI models on AMD hardware alongside its Nvidia stack.
And this work will make OpenAI more agile, allowing it to adopt AMD and potentially other hardware, perhaps even its own in the same timeframe as it begins deploying AMD. That could be awkward to say the least.
If AMD Wins, Who Loses?
Certainly Nvidia CEO Jensen Huang has had better days, but he also knows that this eventually had to happen. AMD’s GPU roadmap looks solid, and the addition of rack-scale system (Helios) infrastructure was inevitable.
But the real loser is Google, whose TPU AI accelerators have been gaining credibility of late. Google has recently begun installing the TPU-based systems at Tier 2 cloud service providers, a first for the company. Now, with OpenAI’s endorsement, these CSPs will be more willing to deploy AMD GPUs.
What Happens Next?
With this deal, AMD is off to the AI races, but note that OpenAI will delay its deployment until the MI450 and, more importantly, the rack-scale infrastructure it will herald become available in late 2026. I expect CSPs to become more interested in AMD hardware as an alternative to Nvidia, which will likely remain the performance leader next year with the Rubin system shipping in roughly the same time-frame as AMD’s MI450.
The AI world just got even more interesting.
Disclosures: This article expresses the opinions of the author and is not to be taken as advice to purchase from or invest in the companies mentioned. My firm, Cambrian-AI Research, is fortunate to have many semiconductor firms as our clients, including Baya Systems BrainChip, Cadence, Cerebras Systems, D-Matrix, Esperanto, Flex, Groq, IBM, Intel, Micron, NVIDIA, Qualcomm, Graphcore, SImA.ai, Synopsys, Tenstorrent, Ventana Microsystems, and scores of investors. I have no investment positions in any of the companies mentioned in this article. For more information, please visit our website at https://cambrian-AI.com.