Why Anthropic is betting billions on cloud servers

Anthropic is paying to use Akamai's cloud infrastructure. Think of cloud infrastructure as renting space in a remote, massive factory full of high-powered computers instead of building your own. Anthropic will use these resources to run its AI. As part of this deal, Akamai is providing Anthropic with warrants, which give them the right to purchase nonvoting preferred stock in Akamai. If fully triggered by Anthropic’s future spending, these warrants could convert into approximately 5% of Akamai's outstanding shares. This is the largest contract in Akamai's history, signaling that Anthropic is planning for massive, long-term growth in the computing work its systems require.

When you interact with a tool like an AI chatbot, you are sending a request to a remote computer that does the heavy lifting. This involves two main components: specialized chips designed for the complex math of AI, and CPUs, which are general-purpose computer processors that handle the background logic and routine tasks. While much of the recent news about AI has focused on the specialized chips that train new systems, this deal highlights the rising demand for CPUs. As AI becomes more advanced and capable of performing complex tasks independently, it requires these versatile processors to manage the data traffic and execute the instructions that allow the system to operate smoothly for millions of users at once.

This deal illustrates the sheer, staggering scale of the financial commitment required to stay in the AI race. It also marks a shift in the way tech partnerships are structured. Usually, the companies making the chips or providing the cloud storage invest money in the AI labs. Here, the roles are being subtly inverted: by giving Anthropic a path to own a small stake in its business, Akamai is effectively locking in a major customer while sharing the potential upside of that relationship. This moves the conversation away from the hype surrounding the AI software itself and toward the stark reality of the physical utility required to sustain it. For the everyday user, this suggests that the future of AI won't just be determined by who writes the best code, but by who has the most reliable access to the massive, invisible factories of computing power that keep these systems running.