Anthropic is reportedly gearing up for a data center push that would rival the largest infrastructure efforts in tech history. The company has been hiring former Alphabet $GOOGL ( ▲ 0.45% )
executives with deep experience building hyperscale facilities, and reports suggest it may aim to develop as much as 10 gigawatts of AI computing capacity over the coming years.
That kind of scale would put Anthropic squarely in the same league as the biggest cloud and AI players.
Why 10 Gigawatts Is a Huge Deal
To put it in perspective, a single gigawatt data center campus is often compared to the power usage of a small city. Estimates from industry executives suggest a 1-gigawatt AI facility could cost around $50 billion to build, implying that a full 10-gigawatt footprint might approach $500 billion in infrastructure over time.
That figure dwarfs what most startups could dream of financing on their own.
Balancing Ambition With Reality
Despite the eye-popping headline number, Anthropic has reportedly told investors it expects to spend closer to $180 billion on AI computing servers through 2029. That gap suggests the 10-gigawatt vision could be a long-term target involving partnerships, phased construction, or shared facilities rather than a single, near-term build.
The company has already taken early steps, announcing a deal with infrastructure partner Fluidstack to develop its first data center projects in New York and Texas, with plans to invest tens of billions in those sites.
As Anthropic and OpenAI push toward potential public listings, control over physical AI infrastructure is becoming just as important as model performance. The AI race is no longer only about algorithms. It is about power, steel, and who can build the biggest digital factories the fastest.