AI pilot to scale programs stall when organizations build first and look for value later. One of the year’s largest AI infrastructure deals did the opposite. Anthropic committed $11.6 billion over seven years to Akamai for dedicated CPU capacity, on top of a $1.8 billion agreement from May, and Akamai committed its own capex only after the customer had signed. According to analyst Sebastian Barros, that sequence is the most transferable lesson in the deal.
AI pilot to scale timeline showing customer commitment before capex in the Akamai and Anthropic deal
Why an AI Pilot to Scale Plan Stalls
A typical enterprise program buys licenses, launches a platform, trains a few enthusiasts and then asks the business where to use it. Pilots multiply, but none has an owner who has promised a measurable result. Budget is spent against hope rather than committed demand.
Akamai took the opposite route. It signed an initial commitment in May, saw it expand roughly four months later, and only then committed about $5.5 billion to build the capacity. It was not predicting demand. It already had it under contract.
The Price of Certainty
Certainty came with a trade-off. Akamai normally targets about 50 cents of annual recurring revenue per dollar of capex, while this contract works out closer to 31 cents. Anthropic also received warrants that could reach about 5% of Akamai. In return, Akamai gets scale, utilization certainty and a seven-year anchor customer.
Internal AI teams face a similar trade. A narrower, committed use case may look less ambitious than a company-wide platform, but it delivers adoption, measurable results and a case for the next budget.
The Numbers at a Glance
| Metric | Figure |
|---|---|
| New Anthropic commitment to Akamai | $11.6 billion over 7 years |
| Earlier agreement (May 2026) | $1.8 billion |
| Total relationship | ~$13.4 billion (+ possible $9 billion) |
| Akamai planned capex | ~$5.5 billion |
| Typical ARR per dollar of capex (Akamai cloud) | ~50 cents |
| ARR per dollar of capex in this contract | ~31 cents |
| Expected contract run rate by end of 2028 | ~$1.7 billion annually |
| Akamai Cloud Infrastructure Services, Q2 revenue | $99 million |
What This Means Depends on Where You Sit
For a vendor or service provider, the lesson is commercial: secure the anchor customer, then size the investment. For a business leader running an internal AI program, the lesson is organizational: secure the sponsor and the target, then release the budget. The principle is identical in both cases, which is that commitment should come before spending.
What This Means for Your AI Pilot to Scale Strategy
Start with a workload, not a tool. Name the process, the team and the volume before choosing a platform.
Secure a committed sponsor. A business owner should accept a measurable target before spending is approved.
Deliver a first tranche, then expand on evidence. Prove the use case in production, and unlock each extension with results from the previous phase. Every AI pilot to scale decision should rest on evidence, not enthusiasm.
Tie training to a named use case. Teams need to specify workflows, supervise agents, handle exceptions and measure outcomes. Generic awareness sessions rarely produce that capability, which is why our industry-specific AI training is built around real use cases.
Run a three-question test before investing. Who owns the result, and what number will they sign? What is the first deliverable, and by when? What evidence unlocks the next phase?
The Bigger Picture
AI pilot to scale is less a technology problem than an order-of-commitment problem. Akamai’s deal shows that a large AI investment can follow a signed customer instead of leading it. The same discipline applies whether the budget is $5.5 billion or $50,000. Once the outcome is committed, the next step is packaging it as finished work, which we cover in Sell Finished Work, Not AI Tools.
Where VisionStratAI helps: Our AI consulting services help organizations prioritize use cases, define sponsor commitments and structure AI pilot to scale roadmaps before capital is committed. Our AI training programmes build the practical skills teams need to deliver on those commitments.
Related reading
- Sell Finished Work, Not AI Tools
- AI Training for Telecom Companies
- AI Training for Finance Teams
- AI Training for Marketing Teams
Ready to define your first committed use case? Talk to an AI expert.
