AI infrastructure company Vesl AI said on March 27 it had secured Nvidia's latest GPUs, the GB200 and B300, at the same time.

Vesl AI said it is the only GPU cloud provider in South Korea to offer the GB200 and B300 simultaneously. It said the B300 has an architecture suited to inference for large-scale AI models, while the GB200 is suited to training very large models. It said it can handle a range of AI workloads.

It also improved accessibility. It provides a self-service format that allows immediate use of major GPUs such as the A100 and H100, as well as storage and an AI development platform, without separate consultations. Users can use GPUs when needed without contract or approval procedures, which it said can speed up AI development.

On costs, it supports a hyperscaler-grade per-minute billing system and a "smart pausing (automatic temporary pause)" function. It said this reduces wasted idle resources and helps cut total cost of ownership for AI infrastructure. It also said intuitive price disclosure and flexible credit management lowered adoption and operational barriers compared with existing GPU clouds.

It plans to expand supply in stages based on global cloud partnerships.

Vesl AI Chief Executive Ahn Jae-man (안재만) said, "After securing supplies of the GB200 and B300, inquiries from companies in need have been increasing rapidly, and we plan to expand supply in stages based on global partnerships." He added, "As a neo-cloud provider with the broadest GPU lineup in South Korea from the A100 to the B300, we will contribute to strengthening domestic companies' competitiveness in AI infrastructure."