The Fermi team intends to launch the mission by the end of 2029. The trajectory was produced by a system built at Physical Superintelligence, a lab that studies physics with AI and opened for business on the day the mission was announced.
Alpha Centauri is 4.4 light-years away and the trip may last 80,000 years; Voyager 1, flying since 1977, has covered less than 1% of that distance.
For a year the engineers at the Fermi Explorer Mission could not find a way to reach another star. A small solar-powered craft on a $15 million budget kept running into the same wall: more power meant more weight, and more weight meant more fuel.
The system returned a route nobody on the team had proposed, set out in a paper that has yet to go through peer review. The craft would slow down until its orbit swings closer to the sun than Mercury’s, then ignite its engine on each tight pass. There the solar arrays take in light four times as intense, and a short burn at high speed yields more energy than the same burn anywhere else. With the engine confined to those passes, the arrays stay modest and the vehicle light.
Alpha Centauri has been attempted before: Yuri Milner pledged $100 million to Breakthrough Starshot in 2016, and a decade on, nothing has launched. “We didn’t want to do another Breakthrough Starshot,” said Philip Johnston, the Fermi mission’s cofounder and president. “We’re dead set on something actually launching.”
Johnston mentioned the problem on a podcast presented by Alex Wissner-Gross, one of PSI’s cofounders, who offered to put it to the lab’s open-source system, Get Physics Done. It breaks a physics question into smaller pieces and works out which simulations to run, leaning on Anthropic’s Claude and OpenAI’s GPT. The trajectory came back a week later.
Matt Pines, PSI’s cofounder and chief executive, said the system worked largely by itself across three days and a billion tokens; a staff astrophysicist held it to the mission’s requirements and checked the output for mistakes. Pines was most surprised that the model produced an entirely different mission profile, and a creative one. The system still lacks a researcher’s judgment, he said: it cannot reliably tell which questions matter, so it stalls on dead ends.
The mission takes its name from Enrico Fermi, who in 1950 asked why a galaxy of hundreds of billions of mostly older stars has sent no sign of anyone else — either crossing to another star is too hard, or nobody has wanted to. A launch would rule out both, and Johnston is left with harsher explanations: that life like ours is almost unimaginably rare, or that intelligence dies out before it can spread. “One of those reasons could be that once you hit superintelligence, that for some reason is self-destructive,” he said.