AMD’s Anthropic deal shows the AI compute race
5 deals show how Anthropic is locking in compute, cash, and chip supply as AI demand strains its infrastructure.

What does AMD’s up to $5 billion Anthropic deal mean for the AI chip race?
AMD and Anthropic’s deal shows how AI firms are buying compute, cash, and supply at once.
| Item | Deal value | Compute / capacity |
|---|---|---|
| AMD | Up to $5 billion | 2 gigawatts of MI450 GPUs |
| SpaceX | $1.25 billion per month | Colossus 1 data center capacity |
| Amazon | Multi-billion-dollar | Not disclosed |
| Google and Broadcom | Multi-gigawatt | Not disclosed |
1. AMD’s bet on Anthropic
Get the latest AI news in your inbox
Weekly picks of model releases, tools, and deep dives — no spam, unsubscribe anytime.
No spam. Unsubscribe at any time.
Advanced Micro Devices has agreed to invest up to $5 billion in Anthropic while also supplying the chips that power the startup’s next wave of model training and inference. The headline number matters, but the structure matters more: AMD is not just selling hardware, it is tying capital to demand.

Anthropic will deploy 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD Helios rack-scale systems. AMD said the first gigawatt should go live in the first half of next year, which gives the deal a clear deployment timeline instead of a vague partnership promise.
- Investor: AMD
- Target: Anthropic
- Investment: up to $5 billion
- Deployment: 2 gigawatts of MI450 GPUs
- First gigawatt: first half of next year
2. Why AMD wants this kind of deal
AMD is trying to grab more of the AI accelerator market, where Nvidia still dominates. Deals like this give AMD a visible customer, a long runway for chip sales, and a way to show that its hardware can support large-scale AI workloads.
The company has already used similar arrangements to deepen ties elsewhere, including a prior deal with OpenAI. In that case, AMD issued a warrant for up to 160 million shares of common stock, linking chip deployment to stock performance and usage milestones.
- Strategic goal: win a larger share of AI infrastructure spending
- Competitive target: Nvidia
- Related customer: OpenAI
- Earlier structure: stock warrant tied to deployment and share price
3. Anthropic’s compute bill keeps rising
Anthropic has spent much of this year locking in more computing power as demand for Claude and Claude Code grows. The company said in April that infrastructure strain was already hurting reliability and performance, especially during peak periods.

That pressure helps explain why Anthropic has been signing multiple supply deals at once. It is not choosing one partner and waiting, it is building a stack of compute sources to keep pace with usage, model training, and product expansion.
- April message: infrastructure strain affected reliability and performance
- Product growth driver: Claude models and Claude Code
- Business signal: compute demand is outpacing current capacity
4. The other infrastructure deals Anthropic has lined up
The AMD agreement is only the latest in a string of contracts Anthropic has announced this year. In May, the company struck a deal with [SpaceX](https://www.spacex.com/) to use all compute capacity at the Colossus 1 data center in Memphis, Tennessee, and the prospectus said Anthropic will pay $1.25 billion per month through May 2029.
In April, Anthropic also signed a multi-billion-dollar agreement with [Amazon](https://www.amazon.com/) and a multi-gigawatt compute agreement with [Google](https://www.google.com/) and [Broadcom](https://www.broadcom.com/). CNBC also reported that Anthropic was in preliminary talks to lease compute from [Meta](https://about.fb.com/), which would add yet another supplier to the mix.
- SpaceX: all compute at Colossus 1
- Monthly payment: $1.25 billion through May 2029
- Amazon: multi-billion-dollar deal
- Google and Broadcom: multi-gigawatt compute agreement
- Meta: preliminary lease talks
5. What the numbers say about Anthropic’s growth
Anthropic’s spending spree makes more sense when you look at the company’s growth. It said in May that its run-rate revenue topped $47 billion, up from about $10 billion for all of last year. That kind of jump can justify massive infrastructure commitments, even if it also raises the stakes for execution.
The company closed a funding round at a $965 billion valuation in the same month and has confidentially filed for a possible IPO. Taken together, those moves suggest Anthropic is preparing for a much larger business, with compute supply now treated like a core strategic asset.
- Run-rate revenue: $47 billion
- Prior full-year revenue: about $10 billion
- Valuation: $965 billion
- IPO status: confidentially filed with the SEC
How to decide
If you want the clearest read on the AI chip market, start with AMD’s role in this deal: it shows how vendors are pairing hardware sales with financial stakes in customers. If you want the bigger business story, Anthropic’s side is more important, because it shows how fast-growing AI companies are turning compute into a long-term supply chain problem.
For investors, the key question is whether AMD can turn these headline partnerships into durable share gains against Nvidia. For operators, the lesson is simpler: AI growth now depends as much on power, data centers, and chip access as on model quality.
// Related Articles
- [IND]
Sriram Krishnan’s path from product to policy
- [IND]
2026-2027 Common App Prompts, Explained
- [IND]
Agentic AI drew four regulators in one week
- [IND]
AI regulation maps the rules you need
- [IND]
OpenAI Status Shows a Busy July for ChatGPT and Codex
- [IND]
Anthropic’s funding hits $132B in 18 rounds