South Korean neocloud companies are accelerating fundraising to secure graphics processing units (GPUs) and expand cloud infrastructure. As surging demand for artificial intelligence computing draws investor interest, these firms are raising their funding targets or pursuing follow-on rounds. With the proceeds going into supply chain expansion and operating technology development, competition to secure a stable GPU supply is expected to intensify.
AI cloud company Mondrian AI will close a 20 billion won ($14 million) Series B round in the fourth quarter of this year, according to industry sources on the 30th. That is 5 billion won more than the 15 billion won originally targeted. The company raised its target after a stream of inquiries from venture capital firms and other investors drawn to the growth potential of the neocloud business.
Mondrian AI provides GPU infrastructure and a cloud operating platform specialized for AI computing, counting LG AI Research and Korea University among its clients. As its neocloud business has grown rapidly, overall company revenue has risen as well. The company expects revenue of 13 billion won to 14.5 billion won this year, more than double last year's 6 billion won. Cumulative revenue through the third quarter reached about 10.7 billion won, already close to its full-year target of 11.2 billion won. "More than 40% of total company revenue is coming from the neocloud business," Mondrian AI said. "We want to take investment mainly from firms that can create synergy with us."
VESSL AI, which closed its own Series B round earlier this year, is now recruiting additional investors. The current round targets 200 billion won. The company is valued at 500 billion won, and some observers expect it to reach 1 trillion won next year. The business is growing quickly, with revenue projected to rise to about 100 billion won this year from 3 billion won last year. VESSL AI has secured roughly 7,000 GPUs and aims to hold 10,000 by year-end.
Elice Group, which lists on the KOSDAQ in October, is counting on its initial public offering to raise funds. Expectations inside and outside the company run high that its market capitalization will top 1 trillion won. When it began the listing process in May, the market valued the company at 700 billion won to 800 billion won.
These companies are focused on raising cash because the neocloud market is growing rapidly as the spread of generative AI drives explosive demand for high-performance GPUs. Neocloud firms secure GPUs in bulk, supply them to companies on demand and provide environments optimized for AI computing. Citing that competitive edge, research firm Gartner has projected that neocloud providers will account for 20% of an AI cloud market expected to be worth $267 billion in 2030. Against that backdrop, Korean firms are moving to seize the opportunity and take the lead in the business.
In Korea, too, GPU supply is failing to keep pace with demand. Some neocloud companies are finding it difficult even to accept inquiries about on-demand services, which let customers rent GPUs as needed. Supply is already tight enough to meet demand from customers seeking large-scale, long-term GPU use.
As a result, competition among neocloud companies to secure GPUs is expected to intensify. Korean firms plan to put the money they have raised into expanding GPU supply chains and into research and development on GPU interconnection and operation. VESSL AI and Mondrian AI in particular run their cloud businesses with most of their GPUs secured through contracts with outside suppliers and companies. That is why the ability to secure GPUs is seen as a key variable separating winners from losers. Mondrian AI linked up with CoreWeave and Nebius as GPU suppliers this year. VESSL AI secured more than 1,000 B200 units through SK Telecom (017670) in August.
An official in the startup sector said Korean neocloud companies have until now competed mainly on price, offering the lowest rates at about half the cost of general-purpose cloud providers. "Now what will matter is the ability to stably operate GPUs scattered across multiple locations, along with AI services," the official said.
