Daegu Startup Sigma Unveils Shared Cloud That Pools Idle GPUs and NPUs

Sigma, an AI infrastructure company based in Daegu, South Korea, has unveiled AICON, a shared cloud platform that links idle GPU and NPU computing resources. A closed beta is planned for this month, with a full commercial launch set for December.

Universities, research institutes and companies that own high-performance accelerators can offer their unused capacity on the platform and earn revenue. AI developers and small and medium-sized businesses can pick the amount of compute their development environment and models need.

For security, AICON relies on container isolation built on Docker and CUDA, with a design intended to keep outside users away from providers' internal data. Sigma has filed patent applications for its resource allocation and scheduling technology.

The company also designs and builds AI servers, supplies ICT equipment and works on public procurement. It was selected for a Korea Startup Promotion Agency re-challenge program, is housed at the Daegu Creative Economy Innovation Center, and says it has hired a chief technology officer while setting up a corporate research institute.

CEO Lee Ji-a said the platform aims to widen access to AI development infrastructure by connecting underused compute, and to help SMEs and startups cut infrastructure costs. From 2027, Sigma plans to win new public and private demand and build an overseas compute supply network.

Source: 서울신문

Raon Community wrote this article from the facts reported in the original. The original belongs to its author; see the source for full details.