AstroForge Is Putting AI in Command of Its Next Spacecraft

The asteroid-mining startup builds "Solo," a transformer-based autonomous control stack, and plans its first fully autonomous flight in 2027 on a Stoke Space rocket.

Published: 2026-09-23 Category: Quick Take Sources: TechCrunch — AstroForge is putting AI in command of its next spacecraft

The Update

AstroForge, the startup developing technology to mine asteroids, is turning to AI to control its spacecraft. The company has built "Solo," an in-house transformer-based autonomous control stack, and plans to fly its first fully autonomous spacecraft in 2027 on the first rocket launched by Stoke Space — a mission backed by NASA that will gather scientific data about the sun.

The company's move is driven by necessity as much as ambition. NASA missions like OSIRIS-REx, which rendezvoused with an asteroid in 2018, relied on teams of flight controllers coordinating from Earth — 100 operators per eight-hour shift. Startups don't have that headcount. AstroForge, founded in 2022 with $56 million in venture funding, has launched two prototypes, both of which suffered anomalies that prevented most mission objectives. In 2025, its Odin spacecraft was launched into deep space but the company couldn't maintain communication — there are few antennas on Earth big enough, and narrow windows to use them — and ultimately lost control of it.

That failure spurred the rethink: put enough intelligence onboard for the spacecraft to solve its own problems.

Why This Matters

Most spacecraft autonomy still relies on traditional control algorithms, in large part because neural networks are considered too unreliable for safety-critical positioning. The first use of a neural network to control a satellite's orbital positioning happened only last year. AstroForge flying AI in command of a deep-space mission would push that frontier significantly — autonomy beyond Earth orbit, where communication delays and dropouts make human-in-the-loop control impractical.

The economics are the driver. For a well-funded startup, a 100-person flight-control team is not viable; AI autonomy is the only path to affordable, frequent deep-space operations. If it works, it could reshape how commercial missions are built, from asteroid mining to orbital servicing.

The Risks

The bet is high-stakes and grounded in past failure. Odin's loss was a communications and control problem — precisely the kind of scenario where on-board intelligence could have helped. But AI in command of a spacecraft cuts both ways: an autonomous system solving problems is good, an autonomous system that makes the wrong call, with no human able to intervene in time, is a different failure mode. AstroForge's own track record — two lost prototypes — makes the first fully autonomous flight a high-wire act. Success would be a genuine milestone; a third failure would test whether autonomy-via-AI is ready for deep space.

Sources: TechCrunch — AstroForge is putting AI in command of its next spacecraft (Sep 22, 2026).