The electronics inside your phone, your car, and every satellite currently orbiting Earth share one critical weakness: heat. Push them past about 200 degrees Celsius and they start to fail. For decades, that thermal ceiling has been one of the hardest walls in engineering.
A team at the University of Southern California may have just found a way around it.
In a study published on March 26, 2026 in Science, researchers led by Joshua Yang, Arthur B. Freeman Chair Professor at the Ming Hsieh Department of Electrical and Computer Engineering of the USC Viterbi School of Engineering and the USC School of Advanced Computing report a new type of electronic memory device that kept working reliably at 700 degrees Celsius, hotter than molten lava and far beyond anything previously achieved in its class. The device showed no signs of reaching its limit. Seven hundred degrees was simply as hot as their testing equipment could go.
"You may call it a revolution," Yang said. "It is the best high-temperature memory ever demonstrated."
A tiny sandwich of extreme materials
The device is a memristor, a nanoscale component that can both store information and perform computing operations. Think of it as a tiny sandwich: two electrode layers on the outside with a thin ceramic filling in the middle. Jian Zhao, the first author of the paper, built this special memristor using tungsten, the metal with the highest melting point of any element, as the top layer, hafnium oxide ceramic in the middle, and graphene on the bottom. Graphene is a sheet of carbon just one atom thick, the same element as diamond, and like the materials around it, it can withstand enormous heat without breaking down.
The result was a device that held data for over 50 hours at 700 degrees without needing to be refreshed, survived more than one billion switching cycles at that temperature, and ran on just 1.5 volts with an operation speed of tens of nanoseconds.
An accidental discovery
Yang's team was originally trying to build a different kind of device using graphene. It didn't work as expected yet. In the process, they stumbled onto something they hadn't expected at all.
"To be honest, it was by accident, as most discoveries are," Yang said. "If you can predict it, it's usually not surprising, and probably not significant enough."
Digging deeper, the team figured out why it worked. In a conventional device, heat causes the metal atoms in the top electrode to migrate slowly through the ceramic layer until they reach the bottom electrode. When they do, the two sides connect permanently, short-circuiting the device and leaving it stuck in the on state, essentially broken.
Graphene stops that process. Its surface chemistry with tungsten is, as Yang put it, almost like oil and water. Tungsten atoms that drift toward the graphene surface find they cannot take hold. Without anything to anchor them, they migrate away. No anchor, no short circuit, no failure.
The team did not just observe this effect. Using advanced electron microscopy, spectroscopy, and quantum-level computer simulations, they figured out exactly what happens at the atomic interface between graphene and tungsten. That mechanistic understanding, Yang said, is what turns a single lucky result into something generally useful. Other materials with similar surface chemistry to graphene could now be identified and tested, potentially making the device easier to manufacture at industrial scale.
Where it could be used
Space agencies have long been calling for electronics that can function above 500 degrees Celsius, roughly the surface temperature of Venus, which has defeated every lander mission sent there. Today's silicon chips fail at a fraction of that.
"We are now above 700 degrees, and we suspect it will go higher," Yang said.
The potential uses extend well beyond planetary exploration. Deep-earth drilling for geothermal energy requires electronics that can survive in environments where the surrounding rock glows red. Nuclear and fusion energy systems generate intense heat near their control equipment. Even for everyday applications, there is a practical benefit: a device rated for 700 degrees is almost indestructible at the 125-degree peaks that car computers routinely face.
What it means for AI
Beyond memory storage, the device has a second capability that makes it particularly relevant for artificial intelligence. The core operation in almost every AI task, from image recognition to language processing, involves a mathematical calculation called matrix multiplication. Today's digital computers perform it sequentially, step by step, burning through enormous amounts of energy in the process. A memristor does it differently. By exploiting Ohm's Law, where voltage times conductance equals current, the device performs the multiplication physically, in the instant electricity flows through it. The answer is simply the current you measure.
"Over 92 percent of the computing in AI systems like ChatGPT is nothing but matrix multiplication," Yang said. "This type of device can perform that in the most efficient way, orders of magnitude faster and at lower energy."
With three co-authors of the paper (Qiangfei Xia, Miao Hu, and Ning Ge), Yang has already co-founded a startup, TetraMem, that is commercializing room-temperature memristor chips for AI computing. His lab has fully working chips made by Tetramem that students use daily to run machine learning tasks at speeds and efficiencies that conventional hardware cannot match. The high-temperature version reported in this paper could extend that capability to places where conventional chips cannot go, letting a spacecraft, a probe, or an industrial sensor process data on site.
How far away is a real product
He is careful not to oversell how close that future is. Memory alone does not make a complete computer. High-temperature logic circuits will also need to be developed and integrated alongside it, and the current devices were built by hand at sub-microscale in a lab. Scale-up will take time.
"This is the first step," Yang said. "It's still a long way to go. But logically, you can see: now it makes it possible. The missing component has been made."
On the manufacturing side, two of the three materials in the device, tungsten and hafnium oxide, are already standard in semiconductor foundries worldwide. Graphene is newer to the industry, but TSMC and Samsung both have it on their development roadmaps, and it has already been grown at the wafer scale in research settings.
The research was carried out as part of the CONCRETE Center , short for Center of Neuromorphic Computing under Extreme Environments, a multi-university Center of Excellence led by Yang at USC and sponsored by the Air Force Office of Scientific Research and the Air Force Research Laboratory. Key measurements and materials characterization were done in collaboration with Dr. Sabyasachi Ganguli's team in the AFRL Materials Lab in Dayton, Ohio. Theoretical analysis was contributed by USC's computational physics group and collaborators at Kumamoto University in Japan.
For Yang, the fact that Science accepted the paper reflects something larger than one lab's results. "Space exploration has never been so real, so close, and at such a large scale," he said. "This paper represents a critical leap into a much larger, more exciting frontier."
