[Edaily Reporter Yun Junghoon ] Vessel AI, an artificial intelligence (AI) infrastructure startup, announced its vision to become a global NeoCloud provider by securing a computing infrastructure comprising 50,000 GPUs by 2028 and 200,000 GPUs by 2030. Vessel AI aims to achieve 2 trillion won in revenue by securing a computing infrastructure of 200,000 GPUs by 2030, starting with 10,000 GPUs within this year.

Ahn Jae-man, CEO of Vessel AI, explained the rationale behind the business model shift and the company’s aggressive scaling goals during the “Vessel AI NeoCloud Strategy Presentation” held at DreamPlus in Gangnam on the 15th, stating, “What customers are purchasing is not simply physical GPUs, but stable and fully integrated large-scale training and inference computing.”

Target of 2 trillion won in revenue by 2028… Aiming for 10,000 GPUs and 400 billion won in financing by year-end

Vessel AI recorded revenue of 130 billion won based on contracts signed in the first half of 2026 and projected that this year’s annual revenue will reach 250 billion won. The company further announced its goal of achieving 700 billion won in revenue in 2027 and 2 trillion won in 2028.

The company is also accelerating the expansion of its GPU infrastructure, a key driver of revenue growth. By leveraging partner data centers and other resources, it plans to increase its current GPU count—currently at 7,000 units—to 10,000 units by the end of this year, 50,000 units next year, and a total of 200,000 units by 2030, including both self-built and jointly owned systems.

To this end, Vessel AI is embarking on a large-scale fundraising effort. The company is currently securing 400 billion won in GPU financing and, next year, plans to pursue an additional 2 trillion won in GPU financing based on already contracted data center infrastructure, in collaboration with global investment banks (IBs) in the U.S., South Korea, and Singapore, as well as eight domestic securities firms. The total capital expenditures (CAPEX) required through 2028 are estimated to range from 5 trillion to 10 trillion won.

The key to accelerating Vessel AI’s growth will likely hinge on whether the company can smoothly secure the planned amount of funding and quickly convert the acquired GPU resources into actual revenue.

Differentiation Through MLOps Technology… Capturing Silicon Valley’s AI Demand

Vessel AI’s ability to rapidly transition to large-scale infrastructure operations is underpinned by its unrivaled MLOps (Machine Learning Operations) capabilities. Having started as an MLOps platform designed to streamline the development, deployment, and operation of AI models, the company has accumulated extensive expertise over many years in software technologies that optimize distributed GPU cluster resources.

For NeoCloud, operational software technology—which reliably connects thousands to tens of thousands of GPUs without latency and efficiently allocates resources—is just as critical as large-scale capital investment.

Vessel AI has proven its technical capabilities by securing a client base that includes not only major domestic AI model development companies such as Upstage and Motif Technologies but also numerous cutting-edge AI labs and R&D firms in Silicon Valley.

CEO Ahn emphasized, “NeoCloud is not simply a business that stacks GPUs in data centers,” adding, “We must possess a technological moat—the ability to predict failures and optimize resources and inference during large-scale training processes—to be chosen by Silicon Valley companies.”

Vessel AI’s strategy is to actively leverage Korea’s geopolitical advantages, its high-bandwidth memory (HBM) semiconductor supply chain, and the government’s AI infrastructure drive policies to compete on equal footing with overseas neo-cloud companies such as Navius and Coreweek. The company plans to lead the “export of AI computing” by securing global AI computing demand based on domestic data center resources and power.

The company has also firmly established a cooperative framework with major domestic corporations and big tech firms. It is currently collaborating with SKTelecom(017670)on a project involving 1,000 NVIDIA B200 GPUs and is expanding its infrastructure based on the B200 cluster it is building with NHN Cloud. It is also diversifying its partnerships with major domestic telecommunications and IT conglomerates, including #SamsungSDS.

Furthermore, to provide a cost-effective AI inference environment, the company is pushing forward with the construction of an AI factory based on domestically produced NPUs (neural processing units). It has completed pilot testing based on Rebellion’s “Atom” and plans to develop an NPU-specialized data center by 2028 in collaboration with domestic NPU companies such as FuriosaAI, with the aim of maximizing return on investment (ROI).

CEO Ahn stated, “We will aggressively secure 500 MW of data center power capacity in Korea by 2028,” adding, “Based on our technological differentiation and a robust domestic infrastructure ecosystem, we will lead the global NeoCloud market.”