Skip to main content

Engineers at University of Southern California Develop ‘Memristor’ Chip Operating at 700°C

Engineers at the University of Southern California (USC) have developed a new “memristor chip” capable of functioning under extreme temperatures. Published in the journal Science on March 26, 2026, the research reveals that this tiny device can store data and perform complex calculations at temperatures as high as 700 degrees Celsius. Typically, current silicon chips used in smartphones or satellites begin to fail beyond 200 degrees Celsius, but this new chip sets a record by operating flawlessly at temperatures higher than molten lava.

Led by Professor Joshua Yang, the research team utilized specialized materials including tungsten, hafnium oxide, and graphene to create this remarkable chip. Conventional chips face short-circuiting and data loss due to metal atoms migrating under extreme heat. However, the strong graphene layers in this memristor chip completely prevent atomic displacement. Laboratory tests demonstrated that the chip can switch states up to one billion times at 700 degrees Celsius while preserving stored data for over 50 hours.

This breakthrough technology holds great promise for sensors used in spacecraft destined for extremely hot environments such as the surface of Venus, as well as underground sensors enduring scorching rock conditions. Beyond data storage, the chip is a significant advancement for artificial intelligence (AI) applications. AI systems like ChatGPT rely heavily on complex mathematical operations such as matrix multiplication, which typically consume substantial energy in current computers. The new memristor chip can perform these computations instantaneously as electrical current flows, vastly accelerating AI calculations while reducing power consumption.

Although currently in the laboratory stage, the technology is expected to take some time before commercial deployment. However, since the materials involved are already used industrially, there is confidence that large-scale manufacturing in the future will be feasible.