Speaking with TFiR, Danielle Cook of Akamai defined the current state of enterprise AI infrastructure delivery and outlined Akamai’s 2026 platform strategy for making AI inference great at global scale.

WHAT IS AKAMAI’S PLATFORM STRATEGY FOR AI INFERENCE IN 2026?

Cook’s answer is a platform thesis built on three integrated components. The first is Akamai’s managed Kubernetes engine — LKE — which serves as the foundational runtime for AI inference workloads. The second is GPU-backed compute, which provides the raw processing capability that AI inference demands at scale. The third is Akamai’s distributed edge, which positions execution close to users globally. What Cook describes is not three separate infrastructure layers being used in parallel — it is a unified stack designed to function as one: “We have a Kubernetes platform, a managed Kubernetes engine — LKE — we’re backing it by our GPUs, we have our distributed edge, and we’re combining that all to make the developer experience great.”

The outcome is a platform where developers can run any AI model or workload wherever their users are, without encountering the operational drag that typically accompanies distributed, multi-component infrastructure.

Developer Experience as a Platform Outcome

Cook’s framing of developer experience as an explicit goal — not a side effect — is significant. In most infrastructure discussions, developer experience is treated as a downstream benefit of good architecture. Cook positions it as a design target in its own right, on equal footing with performance and scale. The implication is that infrastructure which works but is painful to use will slow AI deployment as surely as infrastructure that fails technically. Removing drag from the developer path to production is part of the platform’s job.

Broader Context: How Akamai’s 2026 Strategy Connects to the Larger Predictions

Akamai’s platform priorities in this clip are the product-level execution of everything Cook covered across her full TFiR interview. Her three 2026 predictions — that experience quality will be directly tied to infrastructure decisions, that AI inference placement will become a primary design choice, and that Kubernetes will achieve product-market fit through AI — all point to the same architectural conclusion: inference must run close to users, on a portable, Kubernetes-based runtime, with the operational complexity absorbed by the platform rather than pushed to developers. LKE, GPUs, and distributed edge are Akamai’s answer to that conclusion.

The challenges Cook identified — centralized architecture failure, day-two operational complexity, and the collapse of request-response models under agentic traffic — are precisely the problems this unified stack is designed to solve. And her actionable advice to enterprise leaders — design for distribution first, classify latency-sensitive workloads early, and standardize on Kubernetes — maps directly to what Akamai has built: a managed, distributed, Kubernetes-native platform that makes those three actions executable without requiring organizations to assemble the infrastructure components themselves. “We want to make AI inference workloads great. We want our customers’ experiences to be amazing.”

Watch the full TFiR interview with Danielle Cook here.