Inside Akamai’s Anthropic Deal: The Conditions Behind $11.6 Billion

Start with the clock, not the total

Akamai’s September filing describes approximately $11.6 billion of commitments under two Anthropic project plans. Each initial term runs for seven years from the start of service. That last phrase matters: signing the agreement and beginning the revenue-producing service are different milestones.

The research implication is straightforward. A total contract value is not a current annual revenue figure, and dividing it evenly across the term would invent a recognition schedule the headline does not establish. Investors need the service commencement, delivery profile and actual accounting disclosures to understand when the commitment becomes reported revenue.

Identify the obligations that make the promise enforceable

The arrangements include delivery and availability requirements and contractual termination provisions. The filing describes circumstances involving uncured material breaches and material outages, among other conditions. These are not evidence that a failure has occurred. They describe what can happen if the agreement is not performed as required.

For a supplier building dedicated capacity, the distinction is important. An attractive order can support investment while leaving the operator responsible for delivering a working service. Reading only the commitment amount misses the mechanism that makes those future receipts conditional.

I would put the expected service start, spending schedule and customer payments on the same timeline. If one slips, the other two do not necessarily move with it. That is where an apparently secure revenue commitment can still require substantial financing.

Read the warrant beside the revenue agreement

The customer also received a warrant tied to the commercial relationship. Its vesting depends on specified payments and additional commitments. The warrant means the agreement can affect the supplier’s ownership as well as its sales.

That is why a complete assessment needs a per-share view. More business may create value, but the value attributable to each existing share depends on what the company earns after serving the customer and how the share count changes. The commercial headline does not supply that answer.

The documents do not justify treating the arrangement as either guaranteed profit or a disguised loss. They provide a more useful set of questions: can Akamai deliver, when will the customer pay, what return will the capacity earn, and what ownership does the customer receive along the way? Those are the details to carry into the next earnings report.

The share arithmetic is easy to misread

The filing’s warrant is expressed in preferred shares, not ordinary shares. Up to 387,051 preferred shares are covered, and each is initially convertible into twenty common shares under the stated terms. The filing gives an as-converted equivalent of 7,741,020 common shares.

Using the preferred-share count alone would therefore understate the potential common-equivalent amount. Treating every covered share as already issued and freely convertible would make the opposite mistake. Vesting, exercise, transfer and conversion each have their own conditions.

The exercise price is also stated per preferred share. Any comparison with an ordinary-share price needs to respect the conversion ratio rather than compare two numbers expressed in different units. This is the sort of detail that makes a short agreement summary inadequate for a per-share valuation.

Vesting is tied to behavior under the relationship

The first tranche, covering 40% of the warrant, vests on the first specified payment under Project Plan 3, subject to conditions. The three remaining tranches each cover 20% and vest successively with additional $3 billion commercial commitments under the master agreement. Exercises require cash payment of the aggregate exercise price.

These are distinct commercial milestones. An investor should not substitute an announcement date for a payment date, or a potential future commitment for an already satisfied condition. The filing describes a mechanism through which the customer may earn additional economic participation as the relationship develops.

That does not make the commercial transaction inherently unattractive. Customer incentives can help secure business. The valuation question is whether the expected benefit of that business compensates shareholders for the costs and potential additional ownership claims. The total contract amount cannot answer that without the rest of the economics.

A useful model needs two schedules and one reconciliation

The first schedule follows delivery: when capacity becomes available, when service obligations begin and when the customer pays. The second follows ownership: when each tranche vests, whether it is exercised and how the resulting securities affect common-equivalent claims.

The reconciliation connects those schedules to cash. What does Akamai spend to deliver the service, what does it collect from the customer, and what cash arrives if the warrant is exercised? The exercise proceeds should not be confused with revenue from providing computing.

This approach does not produce a contract valuation from incomplete public information. It prevents several avoidable mistakes before a valuation is attempted. The document gives enough detail to separate a large commercial opportunity from an unconditional promise of profit—and enough ownership detail to make a purely revenue-based reading incomplete.