China’s new chip stores data with a single electron, breaking AI memory bottleneck
Researchers at Fudan University have developed a single-electron flash memory device that could significantly reduce power consumption for AI models on mobile devices. This breakthrough overcomes previous limitations in signal detection for single-electron storage.
Why it matters
This technology could revolutionize mobile computing by enabling complex AI to run locally on devices with minimal energy, bypassing current memory bottlenecks.
How many electrons does it take to store a single bit of information?
For today’s most advanced dynamic random access memory (DRAM) chips from Samsung and SK Hynix, the answer is roughly 200,000.
That is the price of reliability – a vast reservoir of charge needed to ensure a “1” or “0” does not fade into noise.
But a team at Fudan University in Shanghai has just collapsed that number to its theoretical floor: one.
In a paper published in Science on July 16, microelectronics professor Zhou Peng and his colleagues detailed a working two-dimensional (2D) flash memory that traps and reads a solitary electron at room temperature – a feat long regarded as the holy grail for semiconductors.
The device, which they named Guiyi, lifts a faint, fleeting signal from just tens of millivolts to a robust 0.5 volts, a tenfold improvement over any previous single-electron attempt.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in