Friday, October 23, 2026

Zaram Technology Named Lead Institution for National Project on Ultra-Low-Power AI Semiconductors; Aims to Commercialize On-Chip Learning Technology

Input
2026-10-23 12:00:55
Updated
2026-10-23 12:00:55
Zaram Technology corporate identity

[Financial News] Zaram Technology, a fabless company specializing in system-on-chip (SoC) design, has been selected as the lead R&D institution for the Ministry of Trade, Industry and Energy’s national project, “Development of Ultra-Low-Power AI Semiconductor Technology for Field-Adaptive Edge Learning.”
According to financial investment industry sources on the 23rd, the project, managed by the Korea Institute for Advancement of Technology (KEIT), will run for 40 months, from September 2026 through December 2029. The total R&D budget is approximately 6.1 billion won, of which about 4.6 billion won will come from government funding. Zaram Technology will serve as the lead institution, with the Korea Electronics Technology Institute (KETI), Korea University, and Hoseo University participating as joint research institutions. A company providing in-home care services will also participate as a user company and help demonstrate the technology.
The core of the development is an “on-chip learning” semiconductor that trains and adjusts AI models on the device itself. It goes beyond existing on-device AI, which performs inference using pre-trained models, by enabling real-time adaptation to changing environments based on data collected in the field. It also offers privacy benefits by minimizing the need to transmit data externally.
Zaram Technology aims to use neuromorphic design technology to keep power consumption at 10 mW or less in an always-on environment, where sensors monitor activity around the clock. This is one-fifth of the 50 mW requirement in the government’s project announcement. The company plans to develop the technology for use in compact, battery-powered devices as well.
As the lead institution, Zaram Technology plans to turn its existing foundational neuromorphic technology for on-chip learning into commercial products and bring it to market. By the end of the project, it plans to deliver a complete package comprising an AI semiconductor chip capable of on-chip learning, an evaluation board, a software development kit (SDK), and a reference model.
The company also plans to validate the technology in real-world settings. Together with the user company, it will demonstrate care and safety-monitoring services in at least 10 actual households for at least 12 weeks. Based on the results, it plans to expand the technology’s applications to areas including robotics, unmanned aerial vehicles, and industrial equipment.
A Zaram Technology representative said, “The significance of this project lies in its push to commercialize next-generation semiconductor technology that enables AI to learn autonomously inside devices and adapt to changing environments. We will validate our technological capabilities in real service environments and expand business opportunities into diverse markets, including robotics and industrial equipment.”

[email protected] Choi Doo-seon Reporter