Anthropic to pay Akamai $11.6 billion over seven years in cloud deal

Anthropic has signed a seven-year, $11.6 billion cloud infrastructure agreement with Akamai, according to reporting from multiple outlets. The deal could grow to roughly $20 billion depending on usage, and it includes an equity kicker unusual for enterprise cloud contracts: Akamai is offering Anthropic a stake of up to 5% of its stock, with the percentage increasing as Anthropic's spending climbs.

This is not your standard hyperscaler arrangement. Akamai, best known for content delivery and edge computing, is making a CPU-heavy bet at a time when most frontier labs are racing to secure GPU capacity. That positioning suggests Anthropic sees value in workloads beyond pure model training, possibly inference at scale, fine-tuning, or hybrid architectures where CPUs handle orchestration and pre-processing while GPUs focus on the core compute.

The equity component is the real tell. Cloud providers have offered discounts and credits to land marquee AI customers before, but handing over ownership is a different level of commitment. It aligns Akamai's incentives directly with Anthropic's growth and signals how competitive the race to host frontier AI infrastructure has become. For Akamai, it is a chance to reposition from edge specialist to AI infrastructure player. For Anthropic, it is leverage and optionality outside the AWS, Google, Microsoft triopoly.

The size of the commitment also reflects the economic reality of training and serving state-of-the-art models. $11.6 billion over seven years works out to about $1.6 billion annually, which is in line with what top-tier labs are spending on compute as models scale and user bases grow. If the deal hits $20 billion, that is nearly $3 billion per year, a figure that only makes sense if Anthropic expects Claude usage to explode or plans to train significantly larger models.

One detail worth watching is how Anthropic splits workloads across providers. The company already has partnerships with AWS and Google Cloud. This Akamai deal does not replace those relationships, it adds to them. That multi-cloud strategy gives Anthropic redundancy, negotiating power, and the ability to optimize costs by matching workload types to infrastructure strengths. It also means Akamai needs to prove it can compete on performance and reliability, not just price.

The structure here, equity for spend, could become a template if other second-tier cloud providers want to compete for AI workloads. Oracle, IBM, and others are watching. If Akamai can deliver and Anthropic's bet pays off, expect more creative deal structures as AI labs look for alternatives to the big three and cloud providers look for ways to get a seat at the table.

What this means for you: if you are building AI products or evaluating where to run inference workloads, the landscape is shifting. You are no longer limited to AWS, Azure, or GCP. Akamai's CPU-focused infrastructure could be a fit for high-throughput inference, API orchestration, or hybrid pipelines where you need fast, distributed compute without GPU costs. If you are running a customer-facing AI product, try this prompt with your AI assistant: "Help me design a multi-cloud inference architecture that uses CPUs for request routing and preprocessing, and GPUs only for the model forward pass. Show me how to estimate cost differences between a single hyperscaler and a hybrid setup." That exercise will clarify whether deals like Anthropic's make sense for your scale and workload mix.