Anthropic’s $11.6bn Akamai commitment highlights the scale of AI infrastructure demand

Akamai has announced approximately $11.6bn in contractual commitments from Anthropic over seven years. Its distributed cloud infrastructure and software will support Anthropic’s growing CPU workloads, highlighting how AI infrastructure demand extends beyond GPUs.

The scale is significant against Akamai’s existing business. The company generated $4.208bn in revenue in 2025. The Anthropic commitment is approximately 2.8 times that annual revenue, spread over seven years. A simple annual average of $1.66bn would represent around 39% of Akamai’s 2025 revenue, although actual revenue will follow the delivery schedule and contractual terms.

The agreement covers dedicated computing capacity and managed support. Two project plans were signed on 18 September, with each seven-year term beginning when its services start. Payments depend on delivery and service availability, and the arrangements include termination rights. The headline commitment therefore represents future services rather than revenue already earned.

The relationship also includes an equity incentive. Akamai issued Anthropic a warrant for preferred shares equivalent, on conversion, to approximately 7.7 million common shares, or around 5% of its outstanding common stock. The exercise price equates to $111.33 per common share. An initial tranche representing approximately 2% vests upon the first payment under the relevant project plan, subject to conditions; further vesting depends on additional commercial commitments.

Delivering the capacity will require substantial investment. Akamai estimates approximately $5.5bn in related capital expenditure. The relationship also provides scope for up to $9bn in additional commitments, creating a potential route to further expansion.

For Bolt, the announcement reinforces an important aspect of the investment thesis behind CUDO: major AI businesses are making substantial, long-term commitments to external infrastructure providers. Such agreements can give providers greater visibility over future demand and support decisions to invest in new capacity.

The focus on CPU workloads adds another dimension. Our view is that the opportunity extends across the computing resources and supporting infrastructure required to bring AI into everyday use, alongside the GPUs used to train and run models.

For those following CUDO, this agreement provides a significant example of the scale and duration of demand emerging across the sector. Providers that can secure resources, deploy capacity and deliver reliable services have a meaningful role in that expansion, with commercial success ultimately determined by turning customer commitments into profitable delivery.