Anthropic’s $11.6B AI Bet: Why Is Akamai Suddenly in the AI Infrastructure Race?

Akamai has landed a customer commitment big enough to change its future. The harder question is what shareholders must pay to turn that promise into profit. Anthropic’s cloud agreement gives Akamai a much larger role in AI infrastructure but the investment required makes this a test of capital discipline as much as technological ambition.

Let's break down what the Anthropic deal changes for Akamai stock, how its economics could work and what investors should watch in earnings, valuation and cash flow.

What is the Akamai–Anthropic deal?

Akamai announced the agreement on September 24. Anthropic has committed $11.6 billion over seven years for cloud infrastructure supporting CPU workloads. Potential additional business of up to $9 billion is separate from that commitment.

The announcement describes the potential relationship as approximately $20 billion. Straight addition gives $20.6 billion but neither figure should be presented as fully contracted business. The initial commitment is substantial enough without inflating it with unsigned expansion opportunities.

For readers following Akamai stock, the distinction is important: this is a customer purchasing services over time. It is not Anthropic handing Akamai the headline amount in immediate cash or investing that amount in its equity.

Reuters reported a roughly 22% rise in extended trading after the announcement. That reaction shows how quickly investors reassessed the opportunity but it is not a September 25 regular-session closing return.

Source: Reuters report carried by CNA, September 24, 2026.

Why this AI deal matters beyond Nvidia’s chips

The critical detail is the workload. Akamai is supplying CPU infrastructure, rather than announcing a giant purchase of GPUs for this particular agreement. Management confirmed that distinction on its investor call.

CPUs are general-purpose processors. AI systems also need computing resources around the model itself: running applications, coordinating tasks and handling supporting software. Those examples explain the broader opportunity; Akamai has not publicly itemised Anthropic’s complete workload allocation.

Think of a busy restaurant. The chef may receive the attention but more orders also require more capacity at the counter and in the surrounding operation. A better chef alone does not expand the whole restaurant’s throughput.

The investment implication is that AI spending can create business for infrastructure providers beyond accelerator suppliers. That broadens the opportunity set among US technology stocks. It does not establish that CPUs are replacing Nvidia GPUs or Google TPUs.

Akamai’s opportunity is to become useful enough in this surrounding infrastructure that customers commit for years. The commercial challenge is to earn an adequate return after paying for that capacity.

Source: Akamai’s September 24 investor call. Workload examples and investment interpretation are analytical context.

How much could Anthropic change Akamai’s business?

Akamai entered this announcement with sharply different growth rates across its businesses.

Growth was strongest in the smallest of these revenue categories. Meanwhile, earnings declined despite higher sales. An AI contract can change the growth mix but it does not erase the need to improve profitability across the existing operation.

To understand the scale, compare the expected mature Anthropic revenue contribution with the present business. This is a size comparison rather than a forecast that everything else stays unchanged.

One customer could therefore become large enough to reshape the company’s revenue profile. The same calculation reveals the risk: growth becomes more dependent on a particular customer’s ability to honour a large commitment.

Our assessment is that the deal materially strengthens Akamai’s growth prospects. The investment case must now answer a different question: how much economic value does the company retain from that growth?

Calculations use Akamai’s quarterly results and September investor presentation. Annualising a quarter does not constitute annual guidance.

When will the deal generate revenue and cash flow?

The spending comes first. Revenue builds later.

These are management estimates. The table describes this contract rather than Akamai’s total capital expenditure. It also does not mean every maintenance expense disappears after deployment.

The legal timetable deserves attention. Akamai’s SEC filing says each project plan has an initial seven-year term beginning on its respective service start date. Counting seven years from the announcement date would misrepresent the disclosed structure.

Management described the arrangement as take-or-pay upon delivery. That improves demand visibility once the infrastructure is delivered but does not make the revenue unconditional. The filing ties payments to delivery and availability requirements and describes termination rights for specified breaches, outages and insolvency events.

The cash-flow consequence is straightforward. Purchasing equipment uses cash immediately even though the equipment’s cost enters reported earnings gradually through depreciation. Investors could therefore see a much larger business taking shape while near-term free cash flow deteriorates.

This timing gap is central to valuation. Revenue arriving several years from now is worth less today than the same revenue arriving immediately because capital is tied up while shareholders wait.

Can the contract earn an attractive return on Akamai’s investment?

Subtracting capital expenditure from contract value is not a profit calculation. It leaves out the cost of operating the infrastructure, tax, financing and the time between payment and collection.

A more useful starting point is to test the mature operation under different cost assumptions. The following model uses management’s annual revenue expectation and initial investment estimate. Management suggested a seven-year depreciation assumption on the call but did not disclose a precise contract operating margin.

A simple operating-profit sensitivity

Assume the entire investment is depreciated over seven years with no residual value. That produces annual depreciation of approximately $786 million. Then vary annual cash operating costs, including an allowance for allocated overhead, from 20% to 40% of revenue.

These are author-created scenarios rather than company forecasts. Actual depreciation timing, expenses and overhead allocation will differ. The table deliberately excludes the uncommitted expansion.

The model shows why a large contract can produce very different shareholder outcomes. A ten-percentage-point change in the assumed cost ratio changes annual operating profit by $170 million. Revenue alone cannot tell investors which outcome they are paying for.

Depreciation also matters economically even though it is non-cash in a particular year. Servers were purchased with real money. Ignoring their consumption makes the project look more profitable than it is across its full life.

Why a high cash margin is not enough

In the middle scenario, the operation produces $1.19 billion before depreciation, financing and tax. That can look compelling beside its revenue. However, it comes after a substantial capital outlay.

As a rough capital-efficiency check, apply an illustrative 20% tax rate to operating profit and divide by the original investment:

This is a deliberately conservative comparison against original gross investment. It is neither the project’s internal rate of return nor Akamai’s reported return on invested capital. It ignores phased deployment, working capital, depreciation tax timing and any residual equipment value.

Nevertheless, it makes the right question visible. A strong cash margin can coexist with a modest return on the capital required to produce it. A full valuation needs dated cash flows and the price of financing, not just a headline margin.

Our view: the agreement deserves serious credit for revenue visibility. A premium valuation requires evidence that operating costs and deployment spending leave enough value for shareholders after their capital has been committed.

What does Akamai’s stock valuation imply after the announcement?

A valuation comparison needs consistent earnings assumptions and separate price timestamps.

The premarket snapshot corresponds to 3:16 p.m. IST and was approximately 23.7% above the previous regular close. It is not a completed trading-session price. Calculations use the August earnings guidance as a common reference rather than assuming it captures all subsequent deal-related effects.

Investors had already assigned a substantially higher price to the same near-term earnings reference. That is understandable if they expect a better future business. It also means the announcement’s benefits are no longer entirely absent from the price.

The limitation is particularly important here: this is an adjusted earnings multiple. It should not be described as a GAAP P/E. Stock compensation and other adjustments help explain why reported and adjusted earnings can differ materially.

For valuation, we would separate the existing business from the new project. First assess the earnings and cash generation of security, delivery and the current cloud operation. Then add the present value of the Anthropic project’s future cash flows after subtracting the spending needed to create them. Finally account for financing and dilution.

Applying a premium AI multiple to total future revenue skips most of that work. Equally, dismissing the deal because near-term cash flow weakens would ignore the purpose of the investment. The fair question is whether discounted future cash generation exceeds what shareholders must fund today.

How do financing and Anthropic’s warrants affect shareholders?

Akamai reported $4.616 billion of cash, cash equivalents and marketable securities at June 30. Management also pointed to an available $1 billion revolving facility on the September call. These are liquidity resources rather than a net-cash figure: debt and other obligations still matter. Neither figure means every dollar can be devoted to this project without consequences for the rest of the company.

Using cash reduces interest income. Borrowing introduces interest expense and repayment obligations. Additional equity would spread future earnings across more shares. Investors should watch the financing mix rather than assume contract visibility eliminates funding risk.

Anthropic also received a warrant, which is a right to purchase shares at a specified price subject to its terms.

The instrument involves non-voting convertible preferred stock. It does not establish that Anthropic already owns the full potential stake. The filing makes the initial tranche conditional on the first qualifying payment and other terms.

The investor presentation also says the initial warrant’s fair-value revenue reduction is already reflected in the disclosed commitment. Subtracting it again would double-count that adjustment. Dilution is a separate issue: accounting for the customer incentive in revenue does not eliminate the effect on per-share earnings.

For shareholders, the arrangement exchanges some potential upside for a deeper customer relationship. Whether that is attractive depends on the value of the resulting business after investment costs, rather than the prestige of the customer’s name.

What does the deal mean for Google and the global AI market?

This announcement does not establish that Anthropic is abandoning Google parent Alphabet. Anthropic announced an expanded Google and Broadcom agreement in April 2026 for multiple gigawatts of next-generation TPU capacity expected to begin coming online in 2027. It also described using AWS Trainium, Google TPUs and Nvidia GPUs across its operations.

The most reasonable interpretation is that different suppliers can serve different parts of a growing computing requirement. Akamai’s CPU win can coexist with demand for accelerator capacity elsewhere. The announcement alone cannot quantify incremental Google revenue or prove that Google has lost a corresponding amount of business.

For Alphabet investors, the positive possibility is a larger overall market for AI computing. The competitive risk is that sophisticated customers divide their spending among suppliers and negotiate each workload separately. Growth in the overall market does not guarantee that every provider earns the same return.

Globally, this is a useful distinction between AI adoption and AI investment returns. Customers can use more computing while suppliers compete away part of the benefit through pricing or capital commitments. The companies that retain value will need reliable infrastructure, disciplined contracts and a cost advantage that survives competition.

What should Indian investors watch next?

For Indian investors researching how to invest in US stocks, Akamai offers a specific infrastructure exposure. It should not be treated as interchangeable with the broader S&P 500 or the Nasdaq Composite. A single company carries deployment, financing and customer risks that an index spreads across many businesses.

Currency also changes the rupee outcome. As a purely illustrative calculation, a 10% dollar share-price gain combined with a 5% rise in the rupee value of the dollar produces a 15.5% rupee gain before costs and taxes. If the dollar’s rupee value falls 5% instead, that same share-price gain becomes 4.5%. These are scenarios rather than exchange-rate forecasts.

The deal should not automatically become a recommendation for Indian data-centre, power or IT shares either. A domestic company needs a demonstrable contract, order flow or earnings connection before this announcement can support its valuation.

The most useful monitoring framework is operational:

Our stance is constructive on Akamai’s strategic progress but demanding on the financial proof. Anthropic has given the company a credible route to a much larger cloud business. The next stage is to show that this scale earns an attractive return after equipment costs, financing and dilution.

The headline is about a major AI customer. The enduring investment story is whether Akamai can convert that customer’s spending into durable cash generation for each share.