Radiation-Hardened, Fault-Tolerant And Built To Survive The Harshest Environment In The Universe: JPL Just Started Putting NASA’s Next-Generation AI Processor Through Its Paces
According to NASA’s Jet Propulsion Laboratory, engineers at JPL in Southern California have begun testing a next-generation radiation-hardened processor that is showing performance levels roughly 500 times greater than the chips currently used in operational spacecraft. The project, called NASA’s High Performance Spaceflight Computing program, is developing the processor jointly with Microchip Technology Inc., based in Chandler, Arizona, and testing began in February 2026.
The gap between what that number means in practice and what spacecraft are currently capable of is worth sitting with. The processors flying aboard most of NASA’s active missions today were designed for durability above all else.
They were built to survive the brutal radiation environment of deep space, the extreme temperature swings of planetary orbit, and the electromagnetic particle storms that periodically force spacecraft into safe mode, a kind of emergency shutdown that pauses non-essential systems until engineers on Earth can diagnose and resolve the problem. They do all of that reliably. What they cannot do is think.
The new processor is designed to do both. It is a system-on-a-chip, meaning it combines a central processing unit, computational offloads, advanced networking systems, memory and input and output interfaces into a single compact unit small enough to fit in the palm of a hand.
The same architecture is common in smartphones and tablets. NASA’s version, however, is engineered to function without maintenance or repair for years, potentially billions of miles from Earth.
Why does this matter so much right now? Because the distances involved in deep space exploration make human control increasingly impractical. A signal sent from Earth to a spacecraft approaching Mars takes between three and twenty-two minutes to arrive, depending on the relative positions of the two planets.
Round-trip communication therefore takes between six minutes and forty-four minutes at minimum. If a spacecraft encounters an unexpected obstacle, an unstable landing surface, or a scientific anomaly that demands an immediate decision, waiting for instructions from the ground is not an option. The spacecraft must decide.
Current processors cannot support that level of autonomous decision-making at any meaningful speed or complexity. The new processor is specifically designed to change that. With onboard artificial intelligence, a spacecraft running this chip could analyze incoming sensor data, evaluate options and execute a course of action in real time, all without a signal leaving the vehicle.
Jim Butler, the High Performance Space Computing project manager at JPL, described the testing approach directly. The team is using high-fidelity landing scenarios from actual past NASA missions to simulate the kind of processing demands a spacecraft would face during planetary descent, scenarios that would traditionally require dedicated, power-intensive hardware to handle the volume of sensor data involved. The chip is being put through radiation testing, thermal cycling and mechanical shock tests simultaneously. Early results indicate it is performing as intended across all categories.
The applications reach well beyond autonomous spacecraft. NASA says the processor will eventually support Earth orbiters, planetary rovers, deep space probes and crewed lunar and Mars habitats. The agency envisions it enabling faster onboard scientific analysis, more efficient compression and transmission of data back to Earth, and real-time processing of the enormous volumes of information that missions to the outer planets and deep space generate but currently cannot fully utilize.
There is also a commercial dimension. Microchip Technology has already shared sample chips with defense and commercial aerospace partners, and the company plans to adapt the processor for aviation and automotive applications. The chip was selected through NASA’s Game Changing Development program, with JPL choosing Microchip as a partner in 2022. The company funded its own research and development work on the device, which represents an increasingly common model for NASA technology development public mission objectives, private engineering investment.
The team marked the beginning of testing with a symbolic moment: an email titled “Hello Universe,” a deliberate echo of the “Hello World” messages that programmers write when testing new code for the first time. It is a small gesture. But it points to something larger. The history of computing has been defined by the steady expansion of what machines are capable of understanding and doing independently. NASA is now applying that same trajectory to spacecraft, and the implications for how humanity explores the solar system are significant.
Testing continues for several months. When it concludes, the processor will either qualify for spaceflight or return to the engineering team with data on what needs to change. Either way, the question it is answering is one of the most important in space exploration right now. Can we build hardware that is fast enough, durable enough and intelligent enough to send a machine to the edge of the solar system and trust it to make decisions when we are too far away to help?
Early results suggest the answer is yes. Check out our previous lengthy analysis-driven article here.
