Anthropic is betting heavily on Akamai's infrastructure with an $11.6 billion, seven-year commitment that underscores a critical shift in how AI companies source computing power. The deal could balloon to $20 billion if usage scales as expected, making it one of the largest infrastructure contracts in AI history.
The arrangement deviates from typical vendor relationships. Akamai will grant Anthropic equity stakes worth up to 5% of its stock, with the percentage growing as Anthropic's spending increases. This structure aligns both companies' interests in expansion while giving Anthropic a financial upside if Akamai's valuation climbs.
The commitment reveals Anthropic's infrastructure strategy at a pivotal moment. The AI safety company, backed by investors including Google and Salesforce Ventures, has grown dependent on GPU access for training Claude, its flagship large language model. Rather than relying solely on cloud giants like AWS or Google Cloud, Anthropic is diversifying its infrastructure partnerships. This deal prioritizes CPUs over GPUs, a noteworthy choice as most AI labs have historically favored Nvidia's GPUs for large model training.
CPU-based alternatives have gained traction among cost-conscious AI operators. Companies like Lambda Labs and CoreWeave have capitalized on cheaper compute, though GPU dominance persists for cutting-edge model training. Anthropic's pivot suggests the company believes CPU efficiency can support its operational needs at scale, or that a hybrid approach reduces dependency on GPU supply constraints that plague the industry.
The deal also positions Akamai differently. The content delivery network pioneer has historically focused on edge computing and web performance. This contract transforms Akamai into a serious player in AI infrastructure, competing directly with hyperscalers and specialized providers like CoreWeave and Lambda Labs. For Akamai, Anthropic's anchor tenant status validates its data center expansion and provides guaranteed revenue over seven years.
Anthropic's infrastructure choices matter for the broader AI economy. The company competes directly with OpenAI, which relies on Microsoft's Azure infrastructure, and Google DeepMind, which leverages Google Cloud's TPUs. By securing dedicated CPU capacity from Akamai, Anthropic reduces its exposure to GPU shortages and cloud provider pricing power. The equity stake also hedges Anthropic's costs if it needs to expand further.
The timing is significant. Anthropic recently raised funding at a reported $15 billion valuation, making it one of the most valuable AI startups. With $11.6 billion committed to infrastructure over seven years, the company is spending roughly 77% of that valuation on computing power alone. This underscores the capital intensity of frontier AI development and why infrastructure plays like Akamai command such premium arrangements.
The deal also reflects Anthropic's confidence in sustained demand for Claude. Seven-year commitments imply the company expects continued growth in usage and inference costs. If adoption stalls, Anthropic could face overage penalties or underutilization charges, making this a calculated bet on Claude's market success.
Akamai's equity component sweetens the arrangement for both parties. Anthropic gains upside if Akamai's stock appreciates, offsetting some infrastructure costs through capital gains. Akamai retains optionality if Anthropic becomes a major revenue driver and potentially a strategic partner in the AI infrastructure space.