Anthropic, an artificial intelligence company, has entered into a multibillion-dollar agreement with Akamai Technologies, a Cambridge-based network security and content delivery firm, marking a significant contract for the Boston tech sector. The deal, announced late last week, commits Anthropic to purchasing up to $11.6 billion in services from Akamai over seven years, with an option to extend spending by an additional $9 billion, potentially exceeding $20 billion in total.
While Anthropic did not disclose how much of the investment will directly benefit the Boston area or translate into new jobs, the contract represents the largest single deal in Akamai’s history. The agreement comes amid growing concern in the AI industry regarding the rapid development of powerful AI models, some of which experts warn might be evolving beyond human control. However, this deal focuses less on advancing AI research and more on improving the efficiency and distribution of existing AI technologies.
Akamai’s longstanding expertise lies in content delivery networks (CDNs), a system designed to distribute data globally to reduce server strain and improve user experience. In the early internet era, Akamai’s technology enabled popular online content, such as streaming videos, to be delivered from local servers worldwide, minimizing delays and bandwidth bottlenecks. This method has since become standard for companies like Netflix and Apple.
Anthropic’s current challenge is similar: how to deliver AI services quickly and cost-effectively to users at the network’s edge — the points closest to customers and devices. Industry analysts suggest Anthropic will leverage Akamai’s network to bring AI capabilities nearer to end users, reducing the need to send all requests to large centralized data centers.
Anthropic’s advanced AI model, Claude, with trillions of parameters, requires vast computational resources that are typically housed in massive data centers. For everyday questions that do not require such extensive processing power, it could be more efficient to run simpler AI models on Akamai’s distributed servers. Complex queries would be routed to the more powerful central models when necessary.
Reducing latency is a key aspect of this approach. Millisecond-level responsiveness is critical for “physical AI” applications, such as controlling industrial machinery or robots, where delays can impact safety and operational efficiency. By utilizing Akamai’s edge network, Anthropic aims to bring AI processing closer to these applications, mitigating lag time.
As part of the arrangement, Akamai plans to invest $5.5 billion in upgrading its data centers. The company anticipates these enhancements will add a maximum of 110 megawatts of power consumption, significantly less than the energy demands of large hyperscale data centers. The upgrades will be distributed across numerous existing facilities where Akamai already operates.
The new infrastructure will rely primarily on central processing units (CPUs), which, compared with graphics processing units (GPUs) commonly used for AI model training, are more energy-efficient and cost-effective for AI inference—the process of generating responses to user queries. This architecture aligns with Akamai’s goal of efficiently delivering rapid AI-powered answers to end users.
The deal signals a notable advancement for Boston’s tech industry by positioning a local company at the center of the evolving AI infrastructure landscape, even as the broader AI ecosystem grapples with questions about the pace and control of AI development.