Vessel AI, which offers GPU services similar to cloud computing, is targeting unicorn status—valued at over 1 trillion won—within five years by supplying services to major domestic corporations.
On August 20, CEO An Jae-man of Vessel AI discussed the company's growth during a meeting at a shared office in Gangnam, Seoul. Founded five years ago, Vessel AI started with machine learning operations (MLOps) software for efficient GPU resource management. An noted, "Initially, we sold GPU management software to early-stage AI startups, but many asked, 'Why buy management software without GPUs?' This prompted us to launch a GPU cloud business, creating growth opportunities."
The pivotal moment came at the end of last year. An stated, "Clients began requesting GPUs in quantities of 100 and 1,000 starting last year. Following customer demand, GPU cloud services have now become our main business."
Currently, Vessel AI operates with approximately 7,000 GPUs. The company plans to expand this to about 10,000 by the end of this year and aims to increase it to 50,000 next year. An projected, "If we secure 50,000 GPUs, we could rank among the top 10 global neo-cloud providers."
To acquire large-scale GPUs, Vessel AI collaborates with domestic companies. It is currently working with SK Telecom on a project utilizing over 1,000 B200-class GPUs. Discussions are also underway with various information and communication technology (ICT) firms, including KT, for data center and GPU business partnerships. The company is also pursuing participation in the NVIDIA Cloud Partner (NCP) program.
In South Korea, Vessel AI provides cloud services to companies involved in independent AI foundation model projects, such as Upstage and Motif Technologies. It is also participating in the Ministry of Science and ICT's 'Everyone's AI' initiative as part of a consortium with KT, further expanding its connections with domestic firms.
Recently, the company has been focusing on the U.S. market, targeting frontier AI companies that develop large-scale AI models. An revealed, "Currently, half of our total business revenue comes from the U.S., and we have about 20 local clients waiting for GPU supplies."
The contract amount has also increased to approximately 130 billion won. An emphasized, "Depending on GPU availability, we could achieve a contract amount of 250 billion won by the end of this year, with a maximum target of 300 billion won."
Beyond expanding its scale, Vessel AI aims to grow into a global infrastructure company that supports AI researchers in solving problems. An stated, "Rather than solving a single problem directly, we want to provide infrastructure to AI researchers worldwide to accelerate their research. Helping them solve problems more quickly is Vessel AI's role."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.