AMD, Anthropic sign $5bn AI infrastructure deal

AMD and Anthropic have announced a major strategic agreement that signals confidence in the AI startup's ability to compete with OpenAI and other frontier labs.

According to The Wall Street Journal, AMD will invest up to $5 billion in Anthropic whilst the two companies build a 2GW artificial intelligence infrastructure deployment slated to begin in the first half of 2027. It's AMD's boldest move yet into AI infrastructure ownership.

AMD and Anthropic deal

Rather than simply selling chips to customers, AMD is now investing directly in one of the sector's most valued companies. 

Anthropic, maker of the Claude AI model, recently became the world's most valuable AI startup following a substantial funding round. The investment places AMD alongside other major AI customers like Meta and OpenAI, which already use AMD's MI450 systems.

The timing matters. As competition for AI compute intensifies, chip manufacturers are securing their own customer base through equity stakes. AMD's $5 billion commitment signals the company views Anthropic not as another customer, but as a strategic anchor for its AI infrastructure business.

Two gigawatts of compute power are a lot of compute power and will cover most of the model training and inference operations at Anthropic. The H1 2027 launch time frame allows both parties to get their act together regarding data centres and technical integration before launch.

From the standpoint of Anthropic, the deal is a very important one, as it secures hardware access when they need it the most. Frontier AI laboratories face compute limitations all the time. Access to AMD hardware takes away the biggest bottleneck for Anthropic.

Prediction markets have responded decisively. Pricing reflects 82% confidence that Anthropic will reach a $1.25 trillion valuation by the end of 2026, a massive jump from any current valuation.