ANAFLASH, a startup specializing in edge AI processors, has developed an AI microcontroller unit (MCU) featuring proprietary embedded flash memory using Samsung Foundry’s 28 nm process, enabling highly efficient, cost-effective smart edge devices. The new design leverages ANAFLASH’s Logic-EFLASH technology, which integrates zero-standby-power weight memory using only standard logic devices, reducing energy consumption while maintaining performance.
The collaboration with Samsung Foundry began through ANAFLASH’s wholly owned subsidiary, SEMIBRAIN, which won Samsung’s inaugural fabless challenge contest in 2022. Subsequent government research grants in Korea further supported the development of the Logic-EFLASH platform, aimed at building an AI MCU that tightly couples weight memory with computation for energy-efficient AI inference.
“This achievement demonstrates how AI computation can be accelerated efficiently when tightly coupled with zero-standby-power weight memory,” said Younghee Jeon, Head of Korea Research Center at ANAFLASH. “We will continue accelerating innovation on more advanced process nodes.”
Samsung Electronics emphasized its role in supporting innovative startups like ANAFLASH, highlighting the potential to expand the silicon manufacturing ecosystem. Margaret Han, Executive VP and Head of US Foundry at Samsung, said, “We look forward to continuing our collaboration to advance a cost-effective way to integrate nonvolatile memory in advanced logic processes, unlocking new possibilities in intelligent edge computing.”
The AI MCU prototype also demonstrates 4-bits-per-cell embedded flash, enabling efficient storage of trained AI models. The architecture is set to be presented at the IEEE Asian Solid-State Circuits Conference from November 2-5, 2025, in Daejeon, Korea.
ANAFLASH’s AI MCU is currently being sampled to partners and customers, accompanied by software development packages, signaling the company’s readiness to bring energy-efficient, high-performance AI capabilities to battery-powered smart edge devices globally.
