Bengaluru, Mar 17: Akamai Technologies on Tuesday announced the launch of a global-scale artificial intelligence (AI) grid aimed at enabling distributed inference across its extensive network of edge locations worldwide.
The company said its new system, built on NVIDIA’s AI Grid reference design, integrates advanced AI infrastructure with intelligent workload orchestration to route computing tasks across edge, regional and core networks.
The initiative marks a major step in the evolution of Akamai’s Inference Cloud, as it seeks to move beyond centralised AI processing models toward a distributed architecture capable of balancing latency, cost and performance.
According to the company, the platform will utilise thousands of high-performance GPUs to support real-time and large-scale AI applications, including generative and autonomous systems.
Akamai said its intelligent orchestration layer functions as a real-time broker for AI workloads, dynamically directing tasks to the most suitable compute resources.
This approach is designed to improve efficiency by optimising processing speed, cost per operation and system throughput.
The company highlighted that its network of more than 4,400 edge locations enables faster response times by processing data closer to end users, reducing delays associated with centralised cloud systems.
Industry use cases include real-time gaming interactions, financial services applications such as fraud detection and personalised recommendations, and live media processing for global audiences.
At the same time, heavier AI workloads, including large language models and multi-modal processing, will continue to be handled by centralised high-density computing clusters within Akamai’s cloud infrastructure.
The system is built on enterprise AI software and advanced computing architecture, enabling seamless coordination between edge and core environments while maintaining performance and security standards.
Company officials said the development reflects growing demand for scalable, low-latency AI solutions, particularly as new applications require real-time processing at a global level.
