AI Efficiency Race Accelerates as Korean AI Firms See Edge and Low-Power Openings
Published: · Source: hankyung.com

Key points
- At KIW 2026 in Seoul on May 15, 2026, Korean AI companies said competition is shifting from raw computing power to efficiency, on-device processing and edge AI.
- Use the body and FAQ context before acting on this update.
- Compare with related issues inside the category hub.
At Korea Investment Week (KIW) 2026, held on May 15, 2026, at the Shilla Hotel in Jangchung-dong, Seoul, Korean AI companies preparing for IPOs said the AI market is moving from a focus on computing power to a battle over efficiency. Point2 Technology presented its e-Tube interconnect technology for reducing GPU connection bottlenecks, while Ellice Group proposed modular data centers that take four months from design to operation. DeepX said on-device AI will be necessary for autonomous vehicles and robots that require real-time processing, and Mobilint said the edge AI market is growing at more than 20% annually. Mobilint said its NPU can perform high-end inference using 20 to 25 watts and offers better performance per watt than GPUs. Source: hankyung.com.
Partner picks
Relevant partner links for this story
A lightweight commerce block designed to add monetization without breaking reading flow.
Good fit for Korea-based visitors ready to buy.
View offerWorks well for price-sensitive gadget and desk-tool traffic.
View offerUseful for books, work tools, and international shoppers.
View offerAdvertisement
This module may include affiliate links that earn a commission from qualifying purchases. 글로벌스톡펄스


