TL;DR
AstroForge is replacing a $200 million ground control network with an onboard transformer AI called Solo, betting that autonomous spacecraft are cheaper and more resilient than human-operated ones.
What happened
- AstroForge revealed Solo, an in-house transformer-based autonomous control stack designed to run a spacecraft without ground operators.
- Solo is trained on 2,500 onboard sensors and layers traditional control algorithms, subsystem-specific models, and an overarching intelligence agent.
- The system will fly in shadow mode on DeepSpace-2, launching alongside Intuitive Machines' third moon mission before end of 2026, to validate performance before full deployment.
- Autonomy-1, the first fully autonomous mission, is slated for 2027 aboard the debut rocket from Stoke Space, with NASA backing and a solar science objective.
- CEO Matthew Gialich says he currently plans to fly Autonomy-1 with no Earth-receiving radios, making the AI the sole decision-maker onboard.
Why it matters
- The $200 million cost of building a five-dish global ground network is the direct alternative AstroForge is trying to eliminate, making the AI bet an economic necessity, not just a technical one.
- Lost contact with the 2025 Odin spacecraft proved the fragility of ground-dependent operations in deep space, where antenna windows are narrow and rare.
- Most spacecraft still rely on traditional control algorithms; the first neural-network satellite positioning happened only last year, making Solo a significant leap in autonomy maturity.
- A startup with only $56 million in venture funding cannot staff 100-operator shifts like NASA's OSIRIS-REx mission did, so AI autonomy is the only viable scaling path for commercial deep-space ventures.
- Success would validate transformer models as mission-critical infrastructure for space, opening the door for other commercial operators to cut ground costs dramatically.
What to watch next
- DeepSpace-2 shadow-mode results (late 2026): whether Solo correctly diagnoses and resolves anomalies without intervention will determine if Autonomy-1 proceeds as planned.
- Radio decision on Autonomy-1: Gialich says no Earth-receiving radios, but acknowledges team pressure to add them. The final hardware manifest will signal how much confidence the company actually has in Solo.
- Regulatory and NASA posture: a NASA-backed mission with no ground uplink capability is unprecedented; watch for any agency conditions that force a communications fallback.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.