Akamai Launches Edge AI Inference Platform Beta
The platform enables serverless and dedicated model deployment across 17 cities to support low-latency AI agent operations.
Updated on Oct. 28, 2026 in Artificial Intelligence
Live Poll
Do you trust that security sandboxes can effectively prevent AI agents from accessing sensitive internal data?
Akamai has launched its Inference Service in beta, allowing users to deploy open-weight models across a distributed network. The platform offers serverless and dedicated deployment options using Nvidia Blackwell 6000 GPUs across 17 cities.
Why it matters
The service targets growing demand for localized model control and cost management, bridging the gap between centralized cloud compute and decentralized application needs.
Akamai is powering the beta with Nvidia Blackwell 6000 GPUs deployed in 17 cities, supporting a catalog of 12 open-weight models ranging from 8 billion to 2.8 trillion parameters.
The players
Akamai
A global edge computing and content delivery network provider that secures and accelerates online experiences.
Nvidia
A designer of graphics processing units and AI hardware architectures powering data center computing.
The details
The platform integrates with the AI Grid orchestrator released in March 2026, which dynamically routes inference requests based on latency, cost, intent, and data sovereignty requirements. To address security risks, Akamai also introduced an Agent Sandbox that restricts the outbound connections and system access of AI agents, preventing unauthorized modifications to the host environment.
Timeline
March 2026: Akamai released the AI Grid orchestrator.
October 2026: Akamai launched the Inference Service beta.
The Tech Race
The launch positions Akamai within the escalating competition for infrastructure to support autonomous AI agents. This market is projected to grow from $29.2 billion in 2026 to $65.5 billion in 2027 as companies seek tighter security and lower-latency inference.
Developers can now build and deploy models within the existing Akamai edge network, potentially reducing latency by moving computation closer to users. The platform requires integration with the company's existing edge infrastructure, with specific serverless and dedicated hardware tiers now in beta.
The takeaway
Akamai is betting on the need for highly distributed edge inference as firms seek to secure AI agents against unauthorized system access. Watch for further platform updates and pricing disclosures as the service moves toward general availability.
Further reading
For broader trends in infrastructure, explore the latest Artificial Intelligence developments.
Source note: This article includes information reported by Fierce Network.
Live Poll
Do you trust that security sandboxes can effectively prevent AI agents from accessing sensitive internal data?
