For most of its history, Akamai Technologies was known for the servers that quietly moved web pages and video across the internet’s last mile. On Thursday, the Cambridge, Massachusetts, company announced a contract that recasts it as something closer to a landlord for the artificial-intelligence boom.
Anthropic, the maker of the Claude chatbot, agreed to pay about $11.6 billion over seven years for computing capacity on Akamai Cloud, the largest contract in Akamai’s history. The agreement can expand by as much as $9 billion more, bringing the potential total to about $20 billion, according to the companies and securities filings.
The commitment is aimed at Anthropic’s CPU workloads, the general-purpose server work that runs AI systems day to day rather than the specialized chips used to train them. Akamai said the deal extends a relationship the two companies began formalizing through a master services agreement in May.
The structure of the agreement is unusual. Akamai issued Anthropic a warrant to buy non-voting convertible preferred stock representing up to about 5 percent of Akamai’s common shares, at an exercise price of $111.33 a share. Roughly 2 percent of the stock is tied to the initial $11.6 billion commitment, and the rest vests as the relationship expands, with each additional $3 billion in purchases unlocking about 1 percent.
The warrant gives Anthropic a direct stake in its supplier and aligns the two companies’ incentives over a contract that will run most of a decade, analysts said. It also gives Akamai a way to hold onto a customer whose demand for computing is growing faster than almost any other in the industry.
Akamai plans to spend heavily to keep up. The company estimates about $5.5 billion in capital expenditures tied to the commitment, including a roughly $1.7 billion increase in 2026 spending to secure and pre-purchase critical components such as memory. Akamai said the deal will not change its revenue guidance for this year.
Investors treated the announcement as validation of Akamai’s bet on cloud infrastructure, a market long led by Amazon, Microsoft and Google. Akamai shares jumped about 15 percent in extended trading after the news, according to Reuters.
The contract adds to more than $2.8 billion in multiyear cloud-infrastructure commitments Akamai has disclosed across its customer base this year. The company has spent years trying to turn its edge network, thousands of servers scattered across the globe, into a credible cloud alternative, arguing that its distributed footprint suits AI workloads that need low latency.
Akamai’s move into cloud computing dates to its roughly $900 million acquisition of Linode in 2022, which gave the company a cloud platform in a single transaction. Since then it has built out data-center capacity and chased enterprise customers, pitching itself as an underdog to the three hyperscalers that dominate the market.
The company was founded in 1998 by researchers at MIT, and for two decades its business was accelerating the delivery of content for customers from Apple to Netflix. That history still defines its image in the market, even as the cloud unit has become the fastest-growing part of the business.
The CPU workloads at the center of the deal are a different animal from the graphics-chip demand that has defined the AI boom. Training frontier models consumes the scarce Nvidia chips; running those models for millions of users consumes ordinary servers, a need that is expanding as AI moves from research into daily use.
For Anthropic, the agreement is the latest in a string of capacity deals as it races OpenAI and Google for the computing power behind frontier models. The San Francisco company, founded in 2021 by former OpenAI researchers, has drawn up to $8 billion in commitments from Amazon and billions more from Google, and has signed large cloud contracts to keep its servers fed.
The deal also shows how the AI buildout is reshaping the companies that supply it. A content-delivery firm best known for caching websites is now underwriting billions in spending to host a chatbot maker’s workloads, and taking a piece of its customer’s equity to do it.
The deal is a vote of confidence in the idea that not every AI workload belongs on the biggest clouds. As the cost of AI infrastructure has climbed, the industry has begun to distribute computing more widely, and a second-tier cloud with a global network has found a way to compete for business that once flowed automatically to the market leaders.
It is also a financial statement from Akamai about its own future. The company has spent the past few years repositioning itself away from a maturing content-delivery business and toward the faster-growing economics of cloud computing, a transition that has lifted its spending but not yet fully convinced investors it can match the scale of its rivals.
Whether Akamai can deliver the capacity on schedule is the next test. The company is pre-purchasing components and leaning on suppliers to secure memory, a sign of how tight the market for AI infrastructure has become. For now, the contract gives Akamai something it has long wanted: a marquee customer whose growth could pull the rest of its cloud business along with it.
