Akamai’s $2.5B Raise Says Edge AI Clouds Are Limited by What They Can Build

Rows of powered-down AI server racks in an edge data center, illustrating Akamai's $2.5 billion raise and power-limited AI cloud capacity

TL;DR · 30-second read

The Short Version

Akamai, a company best known for quietly speeding up websites and online video, has reportedly been turning away customers who want computing power for artificial intelligence.

The problem is not a lack of buyers. It is a shortage of electricity and space to run the machines. This month Akamai borrowed $2.5 billion from large investors, partly to pay off older debt and partly to build capacity for customers who have already signed up.

The takeaway: the artificial intelligence boom is now limited less by who wants it than by how fast anyone can build and power it.

Yahoo Finance reported that Akamai Technologies is turning away AI business as a power shortage collides with a $14 billion surge in deals. In an 8-K filed November 12, 2026, Akamai disclosed that it closed a $2.5 billion private offering of convertible senior notes: $1.25 billion at 1.00% due November 15, 2028, and $1.25 billion at 1.00% due November 15, 2031.

Net proceeds were approximately $2,459.7 million. Akamai said it will use them to repay part of its $1.15 billion in 0.375% convertible notes maturing September 1, 2027, to fund capital expenditures for its Cloud Infrastructure Services (CIS) business, and for general corporate purposes. About $284.4 million went to hedging transactions designed to limit shareholder dilution.

Executive Summary

Akamai has raised $2.5 billion in low-coupon convertible debt, meaning bonds that investors can later swap for stock if the share price rises enough. The money serves two jobs: refinancing a 2027 maturity and paying for the build-out of CIS, the company’s cloud computing business that now sells capacity for AI workloads alongside its long-established content delivery and security services.

The pairing matters. Chief Financial Officer Ed McGowan framed the raise around executing on “existing customer commitments” and said that as revenue from “committed Cloud Infrastructure Services contracts ramps,” Akamai expects a significant increase in EBITDA margins (earnings before interest, taxes, depreciation and amortization, a common proxy for operating profitability) and operating cash flow beginning in 2027. A company turning away business while borrowing to deliver contracts it has already signed is describing a supply constraint, not a demand problem.

That has implications beyond Akamai. If a provider of Akamai’s scale is capacity-limited, the power and facilities squeeze that has defined hyperscaler expansion is also setting the revenue ceiling for the next tier of cloud and edge providers.

Supply, Not Demand, Is the Binding Constraint

The core claim in the headline rests on two facts read together. First, Akamai is turning away AI business amid a power shortage. Second, its own documents describe the $2.5 billion raise as funding to deliver, not to find, demand: McGowan said the company is “well positioned to execute on our existing customer commitments” and to fund “the continued expansion” of CIS. The forward-looking-statement language in Akamai’s release names the relevant risks plainly: its “ability to recognize anticipated revenue from existing CIS customer commitments” and “the timing and amount of capital expenditures for Akamai’s CIS business.”

When a company’s stated risk is whether it can build fast enough to recognize revenue it has already contracted, the limiting factor is physical: servers, accelerators, data center space and the electricity to run them. AI inference (running trained models to answer queries) and AI training both demand dense racks of power-hungry hardware, and every rack needs a powered, cooled home before it earns anything. Revenue arrives only when that capacity is energized.

The people affected are concrete. Enterprises looking for AI compute outside the largest clouds face fewer options and longer lead times. Colocation landlords and utilities with deliverable power hold pricing leverage. And Akamai’s own growth trajectory now tracks its build schedule more closely than its sales pipeline.

Why the Power Squeeze Reaches Past the Hyperscalers

Most discussion of AI power shortages focuses on hyperscalers, the handful of giants building multi-billion-dollar campuses. Mid-tier and edge cloud providers face the same grid, often with less bargaining power. Providers that rely on leased capacity compete with much larger tenants for the same limited supply of energized space, and utility interconnection queues (the waiting lists for new grid connections) do not prioritize by who has customers lined up.

Edge providers carry an additional constraint. Their value proposition is proximity: compute placed close to users in many metropolitan markets to cut latency, the delay between a request and a response. That means needing power in many locations at once, frequently in dense urban markets where spare capacity is scarcest. A shortage that might be solved by building one remote campus is harder to solve across dozens of cities.

Akamai turning away business is therefore a data point about the market structure, not only the company. The constraint that capped hyperscaler expansion is now capping revenue one tier down, where providers have historically been able to add capacity incrementally.

Reading the Financing Terms

The terms indicate how investors priced Akamai’s plan. The 2028 notes convert at about $112.41 per share, a 40.0% premium to the $80.29 close on November 9; the 2031 notes convert at about $102.37, a 27.5% premium. Both pay 1.00% interest semiannually, starting May 15, 2027. Akamai will pay cash up to principal on conversion and may settle any excess in cash, stock or both, and it spent about $284.4 million, net of warrant proceeds, on hedges intended to reduce dilution if the stock rises past the conversion prices.

Simple arithmetic bounds the capex budget. Net proceeds of roughly $2,459.7 million, less the $284.4 million hedge cost, leaves about $2,175 million. Even if Akamai used all of that to retire the full $1.15 billion of 2027 notes, which it has said it will only partly do, roughly $1 billion would remain for CIS capital spending and general purposes. The actual CIS allocation is likely larger than that floor.

The bet embedded in the deal is timing. Low-coupon debt is cheap to carry, but its value depends on committed CIS revenue ramping as forecast so that, in McGowan’s words, Akamai can increasingly “self-fund future investments from organic cash flow.” If power or equipment delays push capacity out, the contracted revenue shifts with it while the debt does not.

What Buyers and Investors Should Take From It

For enterprises shopping for AI compute, the message is that capacity at alternatives to the largest clouds is not automatically abundant. Securing committed capacity early, with delivery dates in writing, matters more than headline pricing.

For investors, Akamai’s guidance that EBITDA margins and operating cash flow should rise significantly beginning in 2027, with further improvement in 2028, is the measurable claim to track. It depends on build execution and power availability more than on winning new customers, which makes capex timing and energization milestones the variables to watch in coming quarters.

Background

Akamai Technologies, headquartered in Cambridge, Massachusetts, built its business on content delivery, a network of servers that places copies of websites and video close to users so they load faster, and later expanded into cybersecurity. Its Cloud Infrastructure Services business extends that footprint into general-purpose and AI computing, competing for workloads that customers want to run outside or alongside the largest public clouds.

The broader AI infrastructure market has been defined by a scramble for powered data center capacity. Utility grid connections, electrical equipment and suitable sites have become scarce, and that scarcity has shaped how quickly every cloud provider, large or small, can convert customer demand into revenue.

Sources

Source: Akamai Technologies Turns Away AI Business as Power Shortage Meets $14 Billion Deal Surge (Yahoo Finance), on Akamai declining AI business amid power constraints.

Primary sources: Akamai Technologies Form 8-K filed November 12, 2026; Akamai Announces Closing of $2.5 Billion Offering of Convertible Senior Notes (Exhibit 99.1).