Fermi Explorer Mission finds an AI-plotted route to Alpha Centauri

The nonprofit Fermi Explorer Mission announced that it intends to launch a spacecraft to Alpha Centauri, the nearest star system at 4.4 light-years away, by the end of 2029. Even under the best conditions the craft could take up to 80,000 years to arrive. It will follow a novel trajectory discovered by an AI system built by Physical Superintelligence (PSI), an AI physics research lab that is launching today with $58 million in funding led by Breakthrough Energy, the climate-focused investment group founded by Microsoft cofounder Bill Gates.

The mission is a deliberate answer to an earlier failure. In 2016, investor Yuri Milner announced Breakthrough Starshot, which planned to use powerful lasers to push tiny probes to a fifth of the speed of light, reaching Alpha Centauri within 20 years; Milner pledged $100 million toward a proof of concept, but a decade later nothing has launched. "We didn't want to do another Breakthrough Starshot," says Philip Johnston, the cofounder and president of the Fermi Explorer Mission. "We're dead set on something actually launching." His team chose not to constrain the project to a human lifetime and instead set out simply to find a working route to another star.

The mission, currently funded by individual private donors, is budgeted at just $15 million. The spacecraft will carry at least one kilogram of cargo, including artistic and scientific payloads, messages, and a copy of the Golden Record, the gold-plated disc of Earth sounds and images NASA attached to the two Voyager probes in 1977. The engineering problem is severe: Alpha Centauri sits about 25 trillion miles from Earth, and Voyager 1, one of the fastest objects humans have launched and flying since 1977, has covered less than 1 percent of that distance; at its speed the trip would take more than 70,000 years.

Johnston's team spent a year failing to find a route for a small, solar-powered, $15 million craft, stuck on how to supply enough power without adding weight and fuel demand. Johnston raised the problem on a podcast hosted by Alex Wissner-Gross, a physicist who cofounded PSI. Wissner-Gross offered to run it through Get Physics Done, an open-source PSI system that takes a physics research question, breaks it into smaller tasks, and decides which simulations to run, using AI models including Anthropic's Claude and OpenAI's GPT.

A week later the system returned a trajectory Johnston's team had not considered: according to a paper that has not been peer-reviewed, it combines well-known orbital maneuvers in a new way. The craft would first slow down so its orbit swings in closer to the sun than Mercury; on each close pass it fires its engine while the solar panels receive four times the normal light, so a thrust burst delivered there buys more energy than the same burst elsewhere. Because the engine only needs to run near the sun, the solar panels, and the spacecraft, can stay small.

The system conducted the research mostly on its own for three days, running on a billion tokens, says Matt Pines, PSI's cofounder and CEO. An astrophysicist on PSI's staff kept it aligned with the mission's requirements, asked for a cost analysis and clearer charts, and checked the output for errors. "The fact that it came up with an entirely different mission profile, one that was creative and not one [the Fermi team] had considered, that was the more surprising aspect," says Pines. He adds that the model still lacks a human researcher's judgment and taste: it has no reliable sense of which problems are interesting or which approaches are worth pursuing, so it often gets stuck chasing dead ends.

Johnston does not expect the Fermi probe to be first to Alpha Centauri, since spacecraft technology should keep improving; by his own math, an engine just 20 percent faster than today's, built a thousand years from now, would still beat Fermi's probe there by more than 10,000 years. He frames the mission partly as an answer to the Fermi paradox, the puzzle the physicist Enrico Fermi posed in 1950: with hundreds of billions of stars in the galaxy, many far older than the sun, even slow interstellar travel should have let intelligent life spread across it long ago, so why is there no sign of it? Once a Fermi probe actually launches, Johnston argues, humanity becomes a civilization that can and wants to reach another star, which rules out one explanation and points toward others: that life like ours is exceedingly rare, or that intelligent life is common but tends not to survive long enough to spread.

Key facts

The nonprofit Fermi Explorer Mission plans to launch a spacecraft toward Alpha Centauri by the end of 2029 on a $15 million budget from individual private donors; the trip could take up to 80,000 years.
The trajectory came from Get Physics Done, an open-source AI system from Physical Superintelligence (PSI), which uses models including Anthropic's Claude and OpenAI's GPT; PSI is launching with $58 million led by Breakthrough Energy, Bill Gates's climate investment group.
The route has the craft slow into a solar pass closer than Mercury, firing its engine there where panels get four times the normal light, letting the spacecraft and its panels stay small.
The AI system worked mostly unsupervised for three days on a billion tokens; PSI CEO Matt Pines says it lacks a human researcher's judgment and often gets stuck chasing dead ends.
The project is framed against Breakthrough Starshot, the 2016 laser-propulsion plan backed by a $100 million pledge from Yuri Milner that still has not launched a decade later.

Why it matters

This is a case of an AI system producing a genuinely novel engineering result in physics, not just summarizing or coding: Get Physics Done combined known orbital maneuvers into a trajectory the Fermi team's own engineers had not found after a year of trying, on a budget an order of magnitude smaller than the failed Breakthrough Starshot effort. It is a concrete data point for how far automated research agents can go on a hard physical-sciences problem, and for how a tiny nonprofit can now attempt something that used to require Milner-scale money.

Who it affects

Directly, the Fermi Explorer Mission's team and donors, and PSI as the lab commercializing Get Physics Done. More broadly, it is a signal to other physics and engineering research groups that AI-driven simulation search can turn up usable results, and to AI labs and investors watching Breakthrough Energy back a physics-focused AI lab rather than a pure software company.

How to use it

Get Physics Done is described as open-source software that takes a physics research question, breaks it into subtasks, and chooses which simulations to run. The source gives no pricing, access process, or licensing terms for it, and does not say PSI's $58 million round is earmarked for the Fermi mission specifically; the two are separate organizations that collaborated on this one problem.

How solid is it

The trajectory's validation rests on a paper that has not been peer-reviewed, and an astrophysicist at PSI checked the AI's output and requested a cost analysis rather than an independent third party. The mission itself has not launched and is not funded through PSI's round; the individual private donors backing the $15 million budget are not named, and the target date is more than three years out.

Risks and caveats

Even Pines, PSI's own CEO, says the model lacks a human researcher's judgment and taste and often gets stuck on unproductive lines of exploration, meaning the system needed a human astrophysicist checking in on it during the three-day run. The mission's predecessor, Breakthrough Starshot, pledged $100 million in 2016 and still has not launched, underlining how far an announced trajectory is from an actual flight. And by the mission's own framing, even a successful launch would take up to 80,000 years to reach its destination, with Johnston himself doubting the Fermi probe will be the first human spacecraft to arrive.

“Maybe in the next 50 years, there's some great filter that we do not pass through. That all intelligent civilizations, for some reason, do not pass through.”

— Philip Johnston, cofounder and president of the Fermi Explorer Mission