Akamai has spent years trying to convince investors that it is more than a content-delivery and cybersecurity company.
Anthropic may have just changed that conversation.
Akamai announced a $11.6 billion contractual commitment over seven years to provide distributed cloud infrastructure and software for Anthropic’s growing CPU workloads. The relationship can expand by an additional $9 billion, bringing the potential total commitment to about $20 billion.
The deal is enormous relative to Akamai’s existing cloud business.
It also comes with an unusual equity component: Anthropic can ultimately obtain warrants representing up to roughly 5% of Akamai’s common stock, with vesting tied to the initial commitment and future expansion.
For investors, the deal is not simply “Anthropic rents servers from Akamai.”
It is evidence that the AI infrastructure boom is expanding beyond GPUs and hyperscale training clusters into CPUs, memory, distributed cloud and edge infrastructure.
What the Contract Actually Says
Akamai and Anthropic signed project plans under an existing master services agreement.
Anthropic has committed approximately $11.6 billion across seven-year terms, subject to delivery and service-availability conditions.
Akamai says the relationship could grow by another $9 billion if Anthropic buys additional cloud services.
The company issued Anthropic a warrant that can convert into roughly 7.7 million Akamai shares, equal to as much as approximately 5% of the company’s common stock on an as-converted basis.
The exercise price is $111.33 per share.
About 2% of Akamai’s outstanding shares are tied to the initial commitment.
The remaining roughly 3% would vest as the relationship expands, at about 1% for each additional $3 billion of cloud purchases.
That creates a strong incentive for Anthropic to deepen the relationship if the infrastructure performs well.
The Deal Requires Huge Capital Spending
The contract is valuable, but it is not free revenue.
Akamai expects roughly $5.5 billion of cumulative capital expenditure related to the initial $11.6 billion commitment.
About $1.7 billion of additional capex is expected in 2026 to secure critical supply-chain components, including memory.
The company does not expect the agreement to change its 2026 revenue guidance.
Service is expected to begin ramping in 2027.
Akamai’s presentation indicates roughly $150 million to $300 million of revenue in 2027, a larger ramp in 2028 and full contracted run-rate by the end of 2028.
After that, Akamai expects roughly $1.7 billion of annual revenue recognized through the remainder of the contract.
That timing matters.
The market is receiving a large future revenue commitment, but Akamai must spend heavily before the full economics appear in the income statement.
Why CPU Workloads Matter
The most interesting part of the deal is what Anthropic is buying.
This is not described primarily as frontier-model GPU training capacity.
It is CPU-oriented distributed infrastructure.
AI systems need much more than accelerators.
Agents run code.
They call tools.
They retrieve data.
They coordinate multiple processes.
They need networking, storage, memory and conventional compute around the GPU layer.
As AI moves from training models to running millions of persistent agents, CPU demand can become a much larger part of the infrastructure stack.
That is why the Akamai deal has read-through for companies such as AMD and Intel.
Why AMD Investors Paid Attention
Bank of America said the deal strengthens the case for CPU demand in the agentic AI era.
The firm estimates the server CPU market could grow from roughly $61 billion in 2026 to $211 billion by 2030.
That is an analyst forecast, not a guaranteed market outcome.
Still, the logic is important.
AI infrastructure spending is broadening.
The first phase of the boom centered on Nvidia GPUs.
The next phase increasingly includes CPUs, memory, networking, storage, power and distributed inference.
Retail attention around AMD rose sharply after the Akamai announcement, with Stocktwits showing unusually bullish sentiment and high chatter.
That community activity confirms attention, but it does not establish how much of Akamai’s future deployment will use AMD versus Intel hardware.
Why Akamai Is Strategically Different From a Hyperscaler
Akamai’s network is highly distributed.
The company operates infrastructure across a large number of global locations and historically built its business around bringing content closer to users.
That architecture can be useful for agentic workloads.
Not every AI task needs to run in one enormous centralized data center.
Inference, tool execution and latency-sensitive workflows can benefit from distributed compute.
That gives Akamai a different positioning from traditional hyperscalers and GPU-focused neoclouds.
The question is whether it can turn that architecture into attractive margins at much larger scale.
The Warrant Is Both Alignment and Dilution
Anthropic’s warrant creates strategic alignment.
If the relationship grows and Akamai’s stock performs, Anthropic gains economic upside.
But existing shareholders should not ignore dilution.
Up to roughly 5% of common stock is meaningful.
The right way to evaluate the warrant is not to treat dilution as automatically bad.
It should be compared with the revenue, cash flow and strategic value created by the contract.
If Akamai earns attractive returns on a $20 billion relationship, the dilution can be justified.
If margins disappoint or expansion stalls, the economics look less compelling.
Why AKAM’s Price Reaction Was Volatile
Akamai shares jumped sharply after the announcement and were up strongly in premarket trading.
By Friday’s close, Reuters reported the stock up 3.2%.
That difference between the initial surge and the closing move reflects the market’s debate.
The bull case is obvious: this is transformational revenue visibility.
The risk case is equally obvious: Akamai must commit billions of dollars of capital before receiving the full revenue stream.
Investors are therefore valuing both the size of the contract and the execution burden.
The Main Risks
The first risk is customer concentration.
Anthropic becomes an extremely important cloud customer.
The second is capital intensity.
Akamai must deploy equipment before the revenue fully arrives.
The third is supply-chain execution, including memory and server procurement.
The fourth is margin risk.
A huge contract can still destroy value if pricing does not adequately compensate for depreciation, power, colocation and financing costs.
The fifth is Anthropic itself.
The contract is large enough that investors will watch the AI lab’s growth and financial strength closely.
What to Watch Next
Watch Akamai’s 2027 cloud revenue ramp.
Watch capital expenditures relative to plan.
Watch which CPU and memory suppliers win the deployments.
Watch operating margins in Cloud Infrastructure Services.
Watch whether Anthropic exercises additional purchase options.
And watch whether Akamai signs more large AI customers.
The most important implication of the deal is broader than AKAM.
AI infrastructure spending is moving beyond GPUs.
If agentic workloads scale the way Anthropic expects, CPUs and distributed cloud infrastructure may become one of the next major bottlenecks—and one of the next major sources of revenue in the AI buildout.