AMD and OpenAI Ink Megadeal for 6GW of Future AI Compute

3
AMD Helios Rack Scale Infrastructure Announcement 3 MI400 GPUs
AMD Helios Rack Scale Infrastructure Announcement 3 MI400 GPUs

The big news today is that AMD and OpenAI announced a megadeal for AMD’s future generations of AI compute. The deal covers 6GW of capacity starting in 2026. This deal is significant enough to solidify AMD as a major player in the next generation of AI GPUs.

AMD and OpenAI Ink Megadeal for 6GW of Future AI Compute

Broadly, the deal covers 6GW of AI compute infrastructure, starting with 1GW in 2026. At the same time, NVIDIA will invest $100B in OpenAI in its 10GW NVIDIA deal. Given the timeline of 2026, the 1GW is not “Braggawatts” deal. Supply chains already need to be running today to site the installations, so a firm commitment of that first 1GW of deployment means AMD is selling a lot of GPUs. Assuming that OpenAI and AMD are selling “Helios” racks, then that means this will also be a big driver for volume in AMD’s Venice server CPUs.

AMD Helios Rack Scale Infrastructure Announcement And Verano MI500 2027 Path
AMD Helios Rack Scale Infrastructure Announcement And Verano MI500 2027 Path

Likewise, this might be a breakthrough deal for the AMD-Pensando acquisition on the networking side. If AMD is selling its NICs and DPUs into OpenAI as part of this deal, then that will help with volume on that side, even if the enterprise market outside of a few big customers has not picked up for AMD. On the software side, AMD has some big revenue behind making this succeed, so there should be little question on whether AMD is going to have the revenue to justify building out its software operation.

Helios Metrics
Helios Metrics

The companies say that the first 1GW of AMD MI450 GPUs will be deployed in the second half of 2026.

As part of the agreement, to further align strategic interests, AMD has issued OpenAI a warrant for up to 160 million shares of AMD common stock, structured to vest as specific milestones are achieved. The first tranche vests with the initial 1 gigawatt deployment, with additional tranches vesting as purchases scale up to 6 gigawatts. Vesting is further tied to AMD achieving certain share-price targets and to OpenAI achieving the technical and commercial milestones required to enable AMD deployments at scale.” (Source: AMD and OpenAI)

This is an interesting trend, and will point to how we would expect any AI chip startup to operate in the future. If OpenAI is using equity investments or warrants around large AI accelerator purchases, then it seems like any jump in value of an AI chip startup will be captured by the hyper-scale clients in the future.

Final Words

Looking back, the AMD – ZT Systems deal last year appears to be paying dividends. At the time, AMD said that deal was needed to target large-scale deployments. This is certainly a big one.

It seems like a lot of folks are hung up on the warrants for OpenAI, which strike after conditions like AMD hitting a $600 share price. Putting this deal into perspective, the deal’s revenue impact on AMD (margin is of course, different) is likely to be enormous compared to what it would be to NVIDIA’s revenue. As a result, AMD took a deal that requires AMD’s share price to triple. Another impact that is not being widely discussed is that OpenAI is now invested in AMD’s profitable growth over the next several years. If you owned the stock sub-$200 then AMD at $600 is a big win.

The broader impact, of course, is the signal this sends to the market. AMD GPUs are being deployed at scale, at OpenAI. That will give a lot of confidence in the market that AMD is going to remain a player in the AI GPU space. It also says that AMD has something really good in its roadmap since OpenAI made the selection.

Congratulations to the AMD team. This is huge for the company.

Disclosure: We do not provide investment advice. Please consult investment professionals before making investment decisions.

3 COMMENTS

  1. > The broader impact, of course, is the signal this sends to the market. AMD GPUs are being deployed at scale, at OpenAI. That will give a lot of confidence in the market that AMD is going to remain a player in the AI GPU space. It also says that AMD has something really good in its roadmap since OpenAI made the selection.

    Yep, let’s revisit this statement once OpenAI had a chance to try to deploy and run the notoriously brittle AMD software stack at scale. There a are a very lot of bold announcements these days. Not a lot of those headlines will stick to the wall.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.