Anthropic commits $11.6B to Akamai cloud deal with equity stake on the table

Anthropic has agreed to spend $11.6 billion over seven years on Akamai's cloud infrastructure, the two companies confirmed, in a deal that could expand to roughly $20 billion if the AI developer's compute needs keep climbing. The arrangement pairs a long-term capacity commitment with an unusual financial sweetener: Akamai has granted Anthropic rights to acquire up to 5% of its stock, with the size of that potential stake scaling upward as Anthropic's spending grows.

The structure separates this contract from the plain-vanilla cloud deals that have dominated AI infrastructure news. Instead of simply renting capacity, Anthropic is effectively underwriting part of Akamai's expansion while receiving an equity-linked return on its own commitment. Akamai gets a marquee AI customer and a predictable revenue stream; Anthropic gets guaranteed access to capacity and a financial upside if the relationship deepens.

What makes the deal notable is its silicon mix. Akamai's cloud business leans on CPUs rather than the GPUs that most large language model developers chase for training. Anthropic's commitment suggests it sees a durable role for CPU-based infrastructure, likely for inference, data processing, and the less glamorous but high-volume work of serving models to millions of users. Spreading workloads across chip types is one way to ease pressure on GPU supply chains that have constrained the industry for two years.

The dollar figures put the agreement among the largest AI infrastructure contracts disclosed this year. At $11.6 billion, it rivals the scale of commitments that hyperscalers have announced with model developers, though those deals typically bundle GPU clusters and custom accelerators. Akamai, long known for its content delivery network, has been building out cloud computing services as a second act, and this contract gives that effort its biggest validation to date.

For Anthropic, the deal adds another supplier to a compute portfolio that already spans multiple providers, a hedge against depending too heavily on any single vendor. The equity component also gives the company a reason to keep expanding its spend rather than treat the contract as a ceiling. Whether the full $20 billion materializes depends on how quickly Anthropic's model training and inference demands grow, and whether CPU capacity proves as useful to frontier AI work as the company is betting.

The immediate question is how Akamai will finance the buildout. Committing to serve an AI customer at this scale requires data center space, power, and hardware, and Akamai has not detailed how much of the $11.6 billion it will spend upfront versus over time. Watch for capacity timelines, the specific chip vendors involved, and whether other model developers strike similar equity-for-spend arrangements as competition for compute intensifies.