en.Wedoany.com Reported - Gabia (가비아) has launched a hybrid artificial intelligence (AI) infrastructure configuration that combines physical GPU servers with a cloud environment, aiming to reduce cost burdens for enterprises.
Gabia announced on the 9th the launch of a hybrid cloud configuration linking "GPU Server Hosting" and "Gabia Cloud," designed to separate AI learning, inference, and service operations by work phase, thereby lowering both initial investment and operational costs simultaneously.
Recently, with the expansion of generative AI services, GPU demand has surged, increasing the infrastructure cost burden on enterprises. Directly building high-performance GPUs leads to high initial investment costs, while fully operating in the cloud incurs ongoing rental fees, making cost-efficient infrastructure configuration a new challenge.
To alleviate this burden, Gabia's hybrid configuration assigns high-performance computing tasks (such as AI learning and graphics processing) to physical GPU servers based on the RTX 4090, while continuous service operations and traffic management are handled by cloud servers. The company explained that this approach allows enterprises to leverage high-performance GPUs without large initial investments and flexibly adjust resources based on service demand changes.
Specific usage varies by business needs. For image generation service companies, images can be produced in large batches on physical GPU servers, then migrated to the cloud for stable service delivery. AI model development firms can use physical GPUs only during large-scale learning phases, while inference and ongoing services operate in the cloud, reducing costs. Companies running real-time inference services can also connect GPU servers only during new model development to improve infrastructure efficiency.
To commemorate the launch, Gabia is running a promotion: until September 30, new customers who apply will receive Gabia Cloud credits equivalent to the amount paid for GPU server hosting. This offer targets new clients looking to simultaneously utilize high-performance hardware and the cloud, or those considering reducing infrastructure costs and achieving dual-active operations.
Oh Seok, head of the Gabia Cloud Business Team, stated that this configuration was introduced to help enterprises escape the dilemma of choosing between "building their own or going fully cloud," maximizing efficiency by optimizing resources per work phase. He added that leveraging nearly 30 years of accumulated infrastructure operational capabilities, the company will spare no effort in supporting customers to build a secure and economical AI infrastructure environment without initial entry barriers.
