Brain inspired chip could cut AI energy use by up to 70%
A new device uses brain-inspired design to lower power use and improve efficiency in AI computing systems.
Researchers at the University of Cambridge have developed a nanoelectronic device to reduce energy consumption in AI hardware. The team, led by Dr Babak Bakhit, designed the system to mimic how the human brain processes information.
The device uses a new form of hafnium oxide to create a stable, low-energy memristor. It processes and stores data in the same location, similar to how neurons function in the brain.
To achieve this, the researchers added strontium and titanium to form internal electronic junctions. This allows the device to change resistance smoothly without relying on unstable conductive filaments.
Tests showed the device operates with switching currents up to a million times lower than some conventional technologies. It also demonstrated stable multi-level states required for advanced in-memory computing.
The team said the approach could reduce AI hardware energy use by up to 70%. The findings were published in the journal Science Advances.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!