AstroForge is putting AI in command of its next spacecraft
- ID
- 27268
- Status
- summarized
- Published
- 22 Sep 2026, 11:00 PM
- Fetched
- 22 Sep 2026, 11:43 PM
- Provider
- TechCrunch
- Category
- technology
- Original URL
- https://techcrunch.com/2026/09/22/astroforge-is-putting-ai-in-command-of-its-next-spacecraft/
- Source URL
- https://techcrunch.com/feed/
Summary
- Score
- 4.5
- Created
- 22 Sep 2026, 11:44 PM
- Tags
- Audience
- ai_ml_learnersai_agent_users
What happened
Asteroid-mining startup AstroForge is building 'Solo,' a transformer-based autonomous control stack for its next spacecraft, after losing contact with its 2025 Odin mission due to limited deep-space antenna availability. The company plans to fly its first AI-controlled spacecraft in 2027 on Stoke Space's debut rocket, backed by NASA to gather solar data. Most spacecraft autonomy still relies on traditional control algorithms; the first neural network to control a satellite's orbital positioning happened only last year.
Why it matters
This is a concrete example of deploying transformer models in high-stakes, low-connectivity autonomous systems where human intervention is impossible—a design pattern increasingly relevant to AI agent builders operating in edge or intermittent-connection environments. The fact that neural networks are only just entering spacecraft control (last year for basic positioning) shows how conservative mission-critical autonomy still is, which is useful context for anyone building AI agents for production.
Discussion angle
What does AstroForge's failure-driven pivot to onboard AI tell us about when to push autonomy to the edge versus relying on cloud-based agent orchestration—and how does the conservatism of mission-critical systems contrast with the move-fast approach typical in SaaS AI agents?