The US$1.8B contract between Anthropic and Akamai reveals where AI's next battle is being fought: infrastructure

By Simply Wall St · July 3, 2026.

Akamai Technologies, a company born as a content delivery network (CDN) and reinvented as a cloud and cybersecurity provider, has closed the largest commercial deal in its history: $1.8 billion over seven years to supply cloud infrastructure to Anthropic. In parallel, Akamai has completed the acquisition of LayerX, which specializes in secure enterprise browsers under the Zero Trust paradigm, and has been incorporated into Claroty's technology partner program for microsegmentation of critical networks. These are three pieces of the same board: computing capacity to train and run inference on models, and access control to protect those very data flows.

What matters is not just the figure, but who is signing it. Anthropic has not chosen to expand capacity exclusively with Amazon (its main backer and investor), Google or Microsoft; instead it adds Akamai as an additional provider of distributed infrastructure. It is a sign of a pattern we have been observing in the sector: frontier labs are deliberately diversifying their compute supply chain so as not to depend on a single partner, to reduce the risk of bottlenecks and to gain bargaining power against the hyperscalers. Akamai, with its content delivery network already deployed globally at the edge, offers a proposition distinct from the classic centralized cloud: low latency and broad geographic presence, attributes that are increasingly valuable for serving AI models at a global scale.

It is worth separating here the verifiable facts from the financial interpretation. That the contract exists, that it is worth $1.8 billion and that it runs for seven years are data from the company's own announcement. But the projections circulating in the article—revenue of $5.6 billion and profits of $708.5 million for Akamai in 2029, or a fair value of $159.30 per share with 41% upside potential—belong to a model from an analyst in the Simply Wall St community, not to official company guidance nor to an accomplished fact. They are a projection exercise based on assumptions of 9.3% annual growth, and the article itself notes that other, more conservative analysts assume barely 8% and stagnant profits. Treating them as certainties would be a mistake; treating them as a thermometer of investor enthusiasm for the 'AI infrastructure' narrative is more honest.

The risk that the analysis itself acknowledges—and that we endorse—is customer concentration: a contract of this magnitude with a single customer (Anthropic) may boost revenue in the short term, but it also exposes Akamai to the risk that its pace of capacity deployment fails to match the real pace of Anthropic's adoption, compressing margins if capex gets ahead of demand. It is the same dilemma faced by other large cloud providers with their own AI mega-contracts: building data centers and computing capacity under multi-year contracts is a bet that demand for inference and training will keep growing at the promised pace. When it does not—or does so more slowly—the mismatch between capex and the revenue ramp-up hits the balance sheet directly.

The LayerX and Zero Trust piece adds a layer that connects with a thesis we have been repeating at Zendoric: as AI workloads multiply within companies, the risk surface also grows, and the business of securing access to those tools—at the browser, network and identity level—becomes a market as strategic as compute itself. It is no coincidence that Akamai is simultaneously pursuing the infrastructure contract and the strengthening of its security platform: they are two sides of the same coin, because whoever hosts the models also needs to control who accesses them and how.

Our underlying reading is that these kinds of deals—quiet, technical, without the media shine of a model launch—are in fact the true battlefield of the industry in 2026. The war is no longer waged solely on model-quality benchmarks, but on who controls the physical capacity to train and serve them, and who manages the security perimeter around them. Companies that do not build frontier models, such as Akamai, benefit indirectly from that boom by becoming critical suppliers of the 'plumbing' of the AI ecosystem. In the long term, that diversification of infrastructure is good news: the more competition there is among providers of compute, edge and security, the cheaper and more resilient the infrastructure on which the abundance promised by AI is built will be. In the short term, however, the volatility of these multibillion-dollar contracts—and the dependence on AI demand following the promised curve—will remain the variable that decides who wins and who loses among the providers now racing to ride this wave.