Akamai Expands AI Cloud With Managed Inference and Agent Sandbox

Akamai Inference is a managed service that provides access to dozens of open-weight and custom AI models through OpenAI-compatible application programming interfaces. The company said teams can deploy new AI applications and agents at the edge without rewriting code. At launch, customers can choose from 12 open-weight models ranging from 8 billion to 2.8 trillion parameters, including Nvidia Corp.'s Nemotron 3 Ultra, OpenAI Group PBC's gpt-oss-120b and Google LLC's Gemma 4. Akamai is also offering specialized embedding models for retrieval-augmented generation pipelines, designed to connect the open-weight models with its managed database services so responses can be grounded in enterprise data.

Akamai Agent Sandbox is an isolated execution space for AI agents working autonomously on behalf of users. Inside the sandbox, agents run on Akamai Functions, with their code contained in isolated WebAssembly environments managed by the open-source Spin framework. Akamai said enterprises can enforce granular controls over agent compute, memory, system APIs and outbound network access.

The expansion addresses enterprise demand for flexible serving infrastructure and a way to integrate AI with proprietary and third-party data sources under governance protocols, according to the company. Akamai said organizations also need harder boundaries as autonomous agents take on more software coding responsibilities.

Adam Karon, chief operating officer and general manager of Akamai Cloud, said the company expects inference to become the main focus for most enterprises. "Our customers have already proved that thesis," he said. "They want to build, run and protect AI on the same platform. Akamai Inference and Akamai Sandbox add the inference and execution layers that turn our AI infrastructure into a full AI stack, and our security portfolio protects every layer of it."

Akamai's AI Cloud platform combines compute, cloud storage, networking and security, and offers a choice of graphics processing units and other AI accelerators from more than 4,350 edge locations. The company said the platform is designed to run AI inference closer to end users and the data it needs to reduce latency.

Akamai also said it is expanding its fleet of GPU platforms within AI Cloud. It now offers access to Nvidia Blackwell Ultra clusters, specifically RTX PRO 6000 Blackwell Server Edition GPUs. Each eight-GPU Blackwell Ultra node provides up to 2.3 terabytes of total HBM3e memory, which Akamai said supports long-context reasoning. The company plans to add Nvidia's next-generation Vera Rubin GPU platform in the new year.

Editor's Summary

Akamai added Akamai Inference and Akamai Agent Sandbox to its AI Cloud, giving enterprises managed model serving and isolated agent execution across cloud and edge environments. The company also expanded GPU capacity with Nvidia Blackwell Ultra and plans Vera Rubin support next year. The move places inference and agent execution alongside Akamai's networking and security offerings.