Akamai implements NVIDIA AI Grid at global scale

Akamai has unveiled what it says is the first global grid-scale implementation of the NVIDIA AI Grid reference design, which involves interconnecting a geographically distributed AI infrastructure to work as a unified intelligence platform.

The implementation has involved integrating NVIDIA AI equipment into Akamai’s infrastructure and enabling intelligent workload orchestration across its network as part of an effort to develop a distributed grid for AI inference.

Akamai is rolling out thousands of NVIDIA RTX Pro 6000 Blackwell Server Edition GPUs to enable enterprises to run agentic physical AI at global scale with the responsiveness of local compute capacity.

Akamai COO and GM of Cloud Technology Adam Karon said the move represents a step in the evolution of the company’s Inference Cloud, which the company first introduced late last year.

“AI factories have been purpose-built for training and frontier model workloads — and centralised infrastructure will continue to deliver the best tokenomics for those use cases,” he said. “But real-time video, physical AI and highly concurrent personalised experiences demand inference at the point of contact, not a round trip to a centralised cluster.

“Our AI Grid intelligent orchestration gives AI factories a way to scale inference outward — leveraging the same distributed architecture that revolutionised content delivery to route AI workloads across 4400 locations, at the right cost, at the right time.”

Akamai expects the deployment to be able to radically improve cost per token, time-to-first-token and throughput of AI requests. Enterprises will be able to reduce inference costs by matching workloads to the right compute tier automatically.

Potential applications include AI-driven interactions by gaming studios, personalised fraud detection and marketing recommendations for financial institutions, and real-time dubbing of content by broadcasters, the company said.

Image credit: iStock.com/jullasart somdok