AMD Is Betting Up to $5 Billion on Anthropic and Its Claude AI

Key points

• AMD announced it will invest up to $5 billion in Anthropic, the AI safety company behind the Claude family of chatbots.
• Anthropic will deploy up to 2 gigawatts of AMD Instinct MI450 GPUs, the specialised chips that do the heavy number-crunching AI needs, using AMD's new Helios rack-scale system.
• The two companies plan to have the first gigawatt of that computing capacity running by the first half of 2027.
• AMD will use Anthropic's Claude across its own software development and product work as part of a multi-year engineering deal.

AMD and Anthropic announced a sweeping partnership on Wednesday: up to $5 billion of AMD investment behind the Claude chatbot maker, with Anthropic getting access to computing muscle it badly needs in return.

Training and running a powerful AI system requires enormous numbers of specialised chips called GPUs. Anthropic needs more of them. AMD is selling them. Now AMD is also betting serious money that Anthropic will be worth far more in the future.

Under the deal, Anthropic will deploy up to 2 gigawatts of AMD's Instinct MI450 GPUs inside AMD's Helios rack-scale system, a large pre-configured unit that packs many chips together to work as one. A gigawatt is roughly the output of a large nuclear power station, so this is a substantial commitment on both sides. The first gigawatt is scheduled to go live in the first half of 2027, as The Wall Street Journal reported first.

What does this mean for people who use Claude?

More computing power means Anthropic can train better versions of Claude and serve more users at once, so the chatbot should get faster and more capable over time. You don't need to do anything differently today.

For Anthropic, the deal extends a pattern of locking in infrastructure before capacity gets scarcer. The company has already signed deals with Google, Amazon, Broadcom, SpaceX and TeraWulf, and reports suggest a Meta agreement may follow. We covered the Helios system itself on 22 July, when AMD announced rack-scale shipments to Microsoft, Meta and OpenAI and the real hurdles, price and software gaps, came into focus.

For AMD, the partnership matters beyond the hardware sale. The company's own engineers will work with Claude under a multi-year engineering collaboration, embedding Anthropic's model into software development and product work.

Tom Brown, Anthropic's cofounder and chief compute officer, said in the announcement: "By partnering with AMD across the stack, we are securing the capacity we need and optimising it for training and serving Claude."

The deal's real signal is this: the race to build AI infrastructure isn't just about who has the best model anymore. It's about who can secure enough chips and power to keep running at all. AMD's willingness to write a $5 billion cheque, not just sell hardware, tells you how high the competition has pushed those stakes.