Lablup Files Securities Report, Launches KOSDAQ IPO Process
IT DAILY ·
✦ AI Summary
Lablup said on the 25th that it had filed a securities report with the Financial Services Commission and launched the process for its KOSDAQ IPO.
The total number of shares in the offering is 2.18 million, and the indicative price range is KRW 20,000–23,000.
The institutional book-building period is Sept. 28 to Oct. 2, the general subscription period is Oct. 12 to 13, and NH Investment & Securities is the lead underwriter.
Lablup announced on the 25th that it had filed a securities report with the Financial Services Commission and launched the process for its KOSDAQ initial public offering.
The total number of shares to be offered in the deal is 2.18 million. The indicative price range is KRW 20,000–23,000, and the expected total offering size is KRW 43.6 billion to KRW 50.1 billion. The book-building schedule for institutional investors is Sept. 28 to Oct. 2, and the general subscription period is Oct. 12 to 13. NH Investment & Securities is the lead underwriter.
Founded in 2015, Lablup is a company specializing in AI infrastructure management and operations software that improves the efficiency of using high-performance computing resources such as GPU. Its flagship product, Backend.AI, provides integrated single-platform management for GPU and AI accelerators needed for AI model development, training, deployment and operations, and automatically allocates resources according to each user and task.
Lablup's technological differentiator is its in-house GPU virtualization technology. The technology divides expensive GPU into task units so multiple users can use them simultaneously, improving resource utilization.
The technology has also been validated in large-scale AI infrastructure. Lablup participated as the infrastructure partner to the Upstage consortium in the government's 'Independent AI Foundation Model Project,' supporting operation of a GPU cluster of more than 500 units in the first phase and more than 880 units in the second phase. It then advanced to the third phase, proving its ability to stably operate a GPU environment of more than 1,000 units.
At the same time, Backend.AI is not dependent on specific hardware or operating environments. It can manage heterogeneous accelerators such as NVIDIA, AMD and Intel GPU, as well as domestic and overseas NPU, on a single platform, and supports on-premises, public cloud and hybrid environments. It also supports major operating systems.
Lablup is pursuing an IPO based on recent revenue growth and two consecutive years of profitability. Lablup's sales rose from KRW 3 billion in 2023 to KRW 6.7 billion in 2025, more than doubling in two years. Operating profit remained in the black in 2024 and 2025, marking two straight years of profitability. In July, Lablup received preliminary approval from the Korea Exchange for a KOSDAQ listing review.
Lablup plans to use the funds raised through the listing to advance R&D, expand globally and strengthen infrastructure. CEO Shin Jeong-kyu said the listing would serve as a turning point for concentrated investment in R&D and global sales.
Lablup is moving to respond to the diversification of the AI accelerator ecosystem. In particular, it plans to expand the scope of Backend.AI support in line with the diversification of the AI accelerator ecosystem, including GPU and NPU. CEO Shin Jeong-kyu said the company is shifting from a focus on securing GPU volume to a focus on utilization efficiency.
Lablup is also pushing to accelerate expansion into North America, Japan, Southeast Asia, Europe and the Middle East. CEO Shin Jeong-kyu set a goal of becoming a global leader in the AI infrastructure software sector.
Source: IT DAILY · Kwon Young-seok
Original: https://www.itdaily.kr/news/articleView.html?idxno=241196
References
This article was produced with the help of an automated content generation algorithm.
Source: IT DAILY
View originalThis article was summarized and organized by BizCrush based on the original article from IT DAILY. For exact quotations and full details, please refer to the original article.