Pooling Idle GPUs from PC Cafes to Power AI… A Cloud Platform That Captivated Investors

[For a wider range of company information on startups, visit the Unicorn Factory big data platform 'Data Lab'.]

"The biggest burden for AI service companies is the cost of 'inference infrastructure,' such as servers and GPUs (graphics processing units), which support AI in delivering fast and stable responses. AIEEV built a distributed virtual cloud infrastructure by connecting idle PC resources remaining in households or PC cafes, significantly reducing operating costs."

Cha Woo-jin, Senior Reviewer at Blue Point Partners, explained the rationale behind investing in AIEEV, a distributed AI cloud startup, in these terms. Recently, AIEEV raised 600 million won in Pre-Series A funding from Blue Point Partners, Low Partners, and the AI Angel Club. Its cumulative fundraising amount stands at 21 billion won.

Operating an AI service requires high-performance GPUs capable of processing large amounts of data, as well as data centers equipped with them. However, the price of a single high-performance GPU for an AI data center ranges from approximately 33 million to 53 million won, making it prohibitively expensive to secure thousands of units. The burden of purchasing and operating GPUs is cited as the primary reason why AI companies struggle to scale their services or generate revenue.

Senior Reviewer Cha Woo-jin stated, "Operating an AI service requires significant capital to build or lease a data center, and ongoing maintenance costs such as electricity and cooling continue to accumulate." He added, "As a result, many companies find it difficult to sustain their services in the long term."

AIEEV solved this problem by connecting idle GPU resources scattered across the globe into a single network. For example, the company collaborates with PC cafes nationwide to pool the GPUs of computers not being used by customers via a network and deploy them where needed.

Senior Reviewer Cha said, "AIEEV has established a partnership with a company that manages 4,000 PC cafes across the country." He noted, "Since an average PC cafe houses 70 to 100 PCs, securing resources on the scale of tens of thousands of units is possible even if only some are equipped with high-performance GPUs."

AIEEV developed 'Air Cloud,' a network-based AI cloud platform that connects and operates distributed resources without the need for data centers. According to the company, clients have experienced cost reductions of up to 80% compared to existing commercial clouds.

Existing large-scale clouds are complex to operate and manage, making them difficult for startups to handle easily. Managing multiple GPUs stably requires specialized personnel, which is a significant burden for startups that struggle to recruit such high-level talent.

Senior Reviewer Cha said, "Looking at the Air Cloud demo, it supports running and deploying AI models with just a few clicks, without complex infrastructure configuration or specialized personnel." He added, "The fact that there is no need to separately hire infrastructure management experts will be a major advantage for clients."

Air Cloud not only features a function that automatically scales resources as needed but also strengthens its security design by encrypting communication between nodes, making it difficult even internally to easily access data. Senior Reviewer Cha stated, "This high level of security structure became an important factor in the investment decision."

The ability to implement a platform equipped with such technical stability and security is rooted in the capabilities of AIEEV's founding team, which specializes in distributed computing systems. Park Se-jin, CEO of AIEEV, earned a Ph.D. in system software from the Department of Computer Science at POSTECH and worked as a manager at SK Telecom's R&D Institute in 2016. He is currently serving as an adjunct professor in the Department of Computer Science at Keimyung University.

In addition, the founding team consists of system programming and computer science majors from KAIST, Korea University, and Yonsei University, with veterans who have accumulated an average of more than 15 years of experience at companies such as LG, KT, and Tmax handling core development tasks. Senior Reviewer Cha evaluated AIEEV as "a team with differentiated competitiveness in the AI infrastructure field, which has high technical barriers," adding that it "has a great potential to lead the next-generation AI infrastructure market."

Verification of its technological capabilities is also progressing rapidly. Since its establishment in January 2024, AIEEV filed 10 patents in just one year and secured two commercial clients. Currently, it is conducting PoCs (proof of concepts) with more than 10 companies to accelerate market expansion. Additionally, the company is participating in technology and business verification programs from major IT firms, including 'SHIFT,' an open innovation program run by Blue Point Partners and LG Uplus, Samsung C-Lab, and SK Telecom's AI acceleration program, to demonstrate the commercial viability and competitiveness of its distributed AI cloud.

[Money Today Startup Media Platform Unicorn Factory]