AstroForge to Deploy AI in Command of Spacecraft After Odin Failure

The asteroid mining startup develops "Solo" autonomous control stack, plans first fully autonomous mission in 2027.

edit

By LineZotpaper

Published23 September 2026

Read Time3 min

Sources2 outlets

AstroForge, the asteroid mining startup backed with $56 million in venture funding, is developing an autonomous control system that would put a transformer-based AI model in direct command of its spacecraft, following the loss of its Odin probe in 2025 due to communication difficulties. The company plans to fly its first fully autonomous mission in 2027 on the inaugural rocket from Stoke Space, a mission backed by NASA to gather solar science data.

Founded in 2022, AstroForge has launched two prototype spacecraft, both of which suffered anomalies that prevented them from achieving most of their objectives. In 2025, its Odin spacecraft was launched into deep space, but the company could not gain control due to a shortage of large antennas on Earth and limited communication windows. The experience spurred AstroForge to consider putting sufficient intelligence onboard the spacecraft to solve its own problems.

"Would that have been recoverable with all the data on the spacecraft? I don't know, but I can tell you nothing onboard tried it, and I would love something onboard to try if the spacecraft is unrecoverable at launch," said Matthew Gialich, AstroForge's co-founder and CEO.

The company has developed an autonomous control stack called "Solo," a transformer-based model made in-house. Armand Awad, head of flight software, said the company decided to leverage advances in transformer models driven by frontier labs. The stack includes traditional control algorithms, models trained on test data for specific subsystems like power generation or navigation, and an overall intelligence layer trained on about 2,500 sensors in the spacecraft.

"I'm not saying I'm going to make general spacecraft autonomy or general autonomy for the world," Gialich said. "I'm making a constrained autonomy at a very low sensor input, following the basic training of a transformer model."

In theory, the AI agent will perform tasks like anomaly resolution. Awad imagines it realizing it has lost track of its position in space, correlating a power anomaly to issues in its star tracker, and fixing the problem—"probably turning it on and off in this case."

Most spacecraft autonomy currently depends on traditional control algorithms due to concerns about the unreliability of neural networks. The first use of a neural network to control a satellite's positioning in orbit took place just last year, according to a paper on arXiv.

AstroForge's third vehicle, DeepSpace-2, is set to launch alongside Intuitive Machines' third moon mission, expected by the end of 2026. Solo will fly in "shadow mode" on that vehicle, allowing engineers to test it before the Autonomy-1 mission. Gialich noted the trade-off: building a ground network of antennas would cost around $200 million, versus attempting to reduce communication dependency with an AI model.

"The trade for me is: Do I go build my own ground network, which is going to cost [around] $200 million to put up five dishes around the world and then do operations on it, or do I try to remove it with a model?" he said.

Analysis

Why This Matters

If successful, AstroForge's approach could drastically reduce mission costs for deep-space exploration, removing the need for expensive ground-based communication networks.
It represents a leap in spacecraft autonomy, moving from traditional rule-based algorithms to neural network control, which could enable more flexible and resilient missions.
The outcome will be closely watched by NASA and other space agencies as a proof-of-concept for AI-driven spacecraft that can operate without constant human oversight.

Background

Spacecraft autonomy has historically relied on pre-programmed control algorithms and extensive human teams—NASA's OSIRIS-REx mission had 100 operators per shift. Neural networks have been avoided due to concerns about reliability and unpredictability in mission-critical environments. Only in 2025 was the first neural network used to control a satellite's positioning. AstroForge, founded in 2022 with $56 million in venture funding, aims to mine asteroids. Its first two prototype spacecraft suffered anomalies, and the Odin mission failed due to communication issues, prompting the development of more sophisticated onboard intelligence.

Key Perspectives

AstroForge (CEO Matthew Gialich): The company believes constrained AI autonomy is a viable alternative to building a $200 million ground network. The "Solo" system is designed to handle anomaly resolution autonomously using a transformer model trained on spacecraft sensor data.

Traditional space industry / skeptics: Most existing spacecraft rely on proven control algorithms; neural networks are considered risky for safety-critical operations. The industry will be watching for reliability and whether the AI can handle unexpected scenarios without ground intervention.

NASA and scientific community: NASA is backing the 2027 mission to gather solar data, indicating interest in autonomous systems for cost-effective science. The agency has historically invested in automation but has not deployed transformer-based models for primary control.

What to Watch

Performance of "Solo" in shadow mode aboard DeepSpace-2 (late 2026 launch) – whether it correctly identifies and resolves simulated anomalies.
The Autonomy-1 mission in 2027 on Stoke Space's first rocket – the first full test of AI command without ground intervention.
Whether AstroForge can engineer sufficient reliability in the transformer model to gain regulatory and insurance approval for asteroid mining operations.

Sources

AstroForge is putting AI in command of its next spacecraft — TechCrunch
AstroForge is putting AI in command of its next spacecraft — AI News & Artificial Intelligence | TechCrunch