Mobilint CEO Shin Plans Second Chip Mass Production in H2

News|
|
By Kim Ji-young
||
Mobilint's chip. Photo courtesy of Mobilint - Seoul Economic Daily Technology News from South Korea
Mobilint's chip. Photo courtesy of Mobilint

"We began mass-producing neural processing unit (NPU) chips in the second half of last year. We plan to mass-produce additional chips in the second half of this year."

Shin Dong-ju, CEO of Mobilint, recently met with Seoul Economic Daily at his office in Gangnam, Seoul, and said, "The artificial intelligence (AI) inference market is growing rapidly." He added, "Training is being handled by graphics processing units (GPUs) and inference by NPUs, so demand for NPUs will continue to grow."

Mobilint is an AI semiconductor startup founded in 2019 by Shin, a KAIST alumnus. Through Mobilint, Shin is developing NPUs. While GPUs are widely used to develop AI models, alternatives are needed due to their high cost and power consumption. "Hyperscale GPUs cost tens of millions of won, and edge GPUs cost millions of won," Shin explained. "They also consume hundreds of watts of power." In comparison, NPUs, which require less cost and power, are needed as a complementary solution, he said.

Shin Dong-ju, CEO of Mobilint. Photo courtesy of Mobilint - Seoul Economic Daily Technology News from South Korea
Shin Dong-ju, CEO of Mobilint. Photo courtesy of Mobilint

In particular, as physical AI becomes commercialized, demand for NPUs is bound to grow. "The need for chips that can handle various tasks, from the smallest information technology (IT) devices to humanoid robots and automobiles, is increasing," he said. "They must also work with image processors, camera sensors, earphones, and displays."

Mobilint's flagship chip is "Aries," which has been mass-produced since the second half of last year. Aries targets AI-based surveillance facilities. "Installing AI capabilities in each CCTV would require thousands of on-device chips, but the number of chips can be reduced if they are used only for monitoring in a central server room," Shin said. "Aries consumes only 25 watts (W) of power."

"Regulus," being developed as an on-device chip to be embedded directly in devices, is scheduled for mass production in the second half of this year. It can be used in drones, robots, and CCTV cameras. "It took three years from developing the chip to mass-producing it," Shin said. "Aries was mass-produced by Samsung Electronics (005930.KS), and Regulus will be produced by TSMC."

While industrial demand for NPUs is high, developing them is not easy. Semiconductor development is called a "total art" because it requires not only various hardware components but also software. Even if a chip is developed, it is likely to be ignored by the market if developers find it difficult to learn the software to operate it. Securing cost efficiency, power efficiency (performance per power), compatibility, and stability are all considered challenges. "When NPUs are used alongside other existing chips, they must work well in compatibility with other chips," Shin said. "They must also support Windows and Linux." He added, "Even after successfully mass-producing an NPU at a cost of anywhere from tens of billions of won to hundreds of billions of won, the AI market may change rapidly during that time, causing us to miss the trend. Unlike the usual practice of mass-producing chips after receiving purchase orders, NPUs are mass-produced and sold simultaneously even before orders are received."

Despite these difficulties, Shin believes startups actually have potential in the on-device NPU market. "For NPUs used in data centers, it is important to make a single chip well, but infrastructure investment, such as linking hundreds of thousands of chips, must accompany it," he said. "It is not easy for startups to work on data center applications." Shin added, "Many edge products have not been commercialized due to insufficient hardware processor efficiency and cost effectiveness, and the key is to make this possible."

Mobilint has conducted proofs of concept (PoC) with LG Electronics (066570.KS) and Shinsegae I&C. In February, the company received a total of 3 billion won in investment through the "POSCO DX Corporate Venture Capital (CVC) New Technology Investment Association," funded by POSCO DX and POSCO Technology Investment. "We will focus on enhancing AI capabilities while reducing power and cost," Shin emphasized. "Our goal is to improve the strengths of our products."

Original reporting by Kim Ji-young for Seoul Economic Daily.

AI-translated from Korean. Quotes from foreign sources are based on Korean-language reports and may not reflect exact original wording.

00:0002:16

AI KEY

Preview
Korean Corporate Intelligence HubKOSPI · KOSDAQ · 12 sectors

A live, cap-weighted view of every KOSPI and KOSDAQ sector, with same-day Korean reporting distilled by company — built for foreign investors, correspondents and analysts who need to scan Korea before the next session.

Korea Chaebol Tree

Preview
Families Behind the GroupsKFTC May 2026 · DART filings

An English-first interactive map of Samsung, SK, Hyundai, LG and Lotte — built for foreign investors, correspondents and analysts. Korea translates companies into English. We translate the families behind them.