Neuromorphic semiconductors--the “next-generation brains” of artificial intelligence--are set for their first commercial rollout in South Korea. Designed to mimic the human neural network, these chips process and store data simultaneously, offering ultra-low power consumption. This extreme energy efficiency makes them the ideal engine for on-device AI in robotics, automotive systems, and wearable technology.
EdgeAI announced on May 13 that it has entered into a strategic partnership with MDS Intelligence to spearhead the commercialization of this technology. Under the agreement, EdgeAI will lead the chip development, while MDS Intelligence--a subsidiary of the embedded solutions giant MDS Tech--plans to integrate the neuromorphic processors into its smart metering systems used for remote utility management and digital twin energy solutions.
The production pipeline is already taking shape, with manufacturing slated for TSMC, the world's leading foundry. EdgeAI will design the chips with technical support from ASICLAND, a TSMC Value Chain Alliance (VCA) partner. Wafer production is scheduled to begin late this year, with full-scale mass production expected in early 2025.
This project marks a significant milestone as the first instance of neuromorphic technology moving from the laboratory to the commercial market in Korea. While patent filings for brain-inspired computing have surged globally in the US, China, and Korea, a concrete business model for chip commercialization has remained elusive until now. Industry watchers are closely monitoring how this homegrown hardware will reshape the industrial landscape.
The primary draw of neuromorphic architecture is its ability to run AI locally on devices like robots and cars with just 10% of the power required by traditional semiconductors. Unlike the standard “von Neumann” architecture where the processor and memory are separate, neuromorphic chips integrate the two. Furthermore, they operate only when triggered by data changes--much like human neurons--drastically reducing idle energy waste.
This event-based processing is critical for the Edge AI era, where low-power performance is the deciding factor for products like Apple's Vision Pro, Meta's smart glasses, humanoid robots, and autonomous mobility devices. According to MarketsandMarkets, the global neuromorphic computing market is projected to skyrocket from $28.5 million in 2024 to $1.32 billion by 2030, representing a staggering annual growth rate of 89.7%.
To accelerate development, EdgeAI has partnered with US-based BrainChip, adopting their neuromorphic Intellectual Property (IP). This follows a proven industry model similar to how Samsung or Qualcomm use ARM's IP to build their final SoC (System on Chip) products. EdgeAI intends to combine this IP with its proprietary hardware and software stacks to deliver a highly competitive product.
“We are committed to establishing the first successful commercial use case for a domestic neuromorphic chip,” said Kim Jung-wook, CEO of EdgeAI. “Alongside MDS Intelligence, we will aggressively pursue expansion into the global market.”
Ji Chang-gun, CEO of MDS Intelligence, added, “By combining EdgeAI's innovative technology with our expertise in IoT-based smart metering, we will ensure that Korean semiconductors can compete at the highest level globally.”
Meanwhile, EdgeAI, which is headquartered in Gwangju, opened a new Seoul office on the same day to secure top-tier talent and strengthen its R&D capabilities as it prepares for the upcoming production cycle.
