[Edaily Reporter Kim Hyung-il ] LabUp, a company specializing in AI infrastructure operations software, is set to target the domestic AI software market following its listing on KOSDAQ. The company’s strategy is to capitalize on businesses’ demand for more efficient AI infrastructure operations by leveraging its platform, which supports everything from the management of AI accelerator resources—such as GPUs and NPUs—to model training, inference, and service operations.

LabelUp is set to list on the KOSDAQ market on the 3rd. LabelUp’s core competitive advantage lies in “Backend.AI,” an integrated AI infrastructure operations platform. It provides unified management of a company’s computing resources—such as GPUs and NPUs—and supports AI model training, inference, and service operations. The software goes beyond simply building AI infrastructure; it helps companies efficiently utilize their limited computing resources.

It has also demonstrated tangible improvements in resource utilization efficiency through real-world applications. LabelUp operates its platform at over 120 sites domestically and internationally and has experience managing more than 2,000 GPUs at a single site. In customer case studies, it increased GPU utilization rates by 2.8 times compared to previous levels. LabelUp explained that it has proven its technical capability to efficiently allocate and manage the computing resources required for AI service operations.

The company has also built its revenue structure around a subscription-based business model. This structure generates additional license revenue whenever a client expands its AI infrastructure or broadens the scope of platform usage. Last year, recurring revenue accounted for 94.7% of total revenue. A key feature of this model is the ability to secure consistent license revenue rather than relying on one-time equipment sales or project contracts.

Financial performance continues to show growth. Revenue for the first half of this year reached 4.7 billion won, a 58.2% increase compared to the same period last year. The company maintained profitability with an operating profit of 560 million won. The company explained that while revenue growth is driven by an expanding customer base and the expansion of AI infrastructure by existing customers, it is securing profitability through its recurring-revenue-based business model.

Recently, the company has been expanding its business from managing existing AI infrastructure to managing AI models and token usage. This is because AI companies are facing a growing need to manage costs and usage arising from the actual operation of services, beyond simply adopting models. LabelUp plans to expand its business scope with an integrated platform that encompasses everything from computing resources to model operations.

LabelUp’s business expansion is also aligned with shifts in the global AI investment market. Until now, the AI investment market has focused on building infrastructure such as data centers, GPUs, and memory semiconductors. However, as large-scale capital investments continue, a new challenge has emerged: how to efficiently utilize the secured resources and apply them to actual services. Consequently, the role of software companies that support AI infrastructure operations, model optimization, and service commercialization is growing.

Interest in software companies is also resurging in the global investment market. The “iShares Expanded Technology-Software Sector ETF (IGV),” which invests in U.S. software companies, fell 14.28% in the first half of this year but has risen 22.06% in the second half through the 25th of last month. This stands in contrast to memory semiconductor-related ETFs, which had driven the AI investment boom in the first half of the year but have shown weakness in the second half. Some analysts interpret this as a sign that investment interest is shifting from AI infrastructure expansion to the software sector, which utilizes and operates that infrastructure.

In the domestic stock market as well, AI software companies are emerging as new investment options. Until now, domestic AI-related investments tended to be concentrated on semiconductor companies such as SamsungElectronics(005930)and SK hynix(000660). This was because hardware companies, such as memory semiconductor and GPU manufacturers, were considered the key beneficiaries during the growth of the global AI industry. However, as specialized software companies—focused on AI infrastructure operations and model optimization—are successively entering the stock market, the scope of investment is broadening.

Nota Inc., which listed on KOSDAQ last November, is a company with AI model lightweighting and optimization technologies. Through its proprietary platform, “NetsPresso,” it enables AI models to run efficiently across various hardware environments. The company is expanding its business scope from its existing focus on mobile and edge AI into the data center and enterprise AI markets. In the first half of this year, the company recorded revenue of 7.4 billion won and a backlog of 11.6 billion won.

MakinaRocks Co., Ltd., which debuted on KOSDAQ last May, supports the entire process—from development to deployment and operation—of AI models in industrial settings such as manufacturing and defense, based on its industrial AI operating system “Runway.” A key feature is its ability to utilize AI even in security-critical, closed-network environments. In the first half of this year, revenue reached 7 billion won, a 168% increase year-over-year, and new order intake exceeded 20 billion won.

These companies operate in different segments of the AI industry. While LabUp focuses on AI infrastructure operations and resource management, Nota Inc. specializes in AI model optimization, and MakinaRocks Co., Ltd. excels in industrial field applications. They share a common goal: to reduce costs and operational inefficiencies that arise during the actual utilization phase following the establishment of AI infrastructure.