Key Takeaways:
- AstroForge is pioneering “Solo,” an in-house developed, transformer-based AI control stack, to enable fully autonomous spacecraft operations and overcome the cost and logistical limitations of traditional human-controlled missions.
- Driven by prior mission anomalies, AstroForge recognized the critical need for onboard intelligence, choosing to invest in AI to resolve deep space challenges autonomously rather than building an expensive global ground communication network.
- The company plans a bold “Autonomy-1” mission in 2027, potentially flying with minimal Earth communication, following a “shadow mode” test on the DeepSpace-2 mission in 2026, marking a significant leap towards self-sufficient, commercial space exploration.
Fly to an asteroid, land on it, make no mistakes: If only it were that easy. The immense challenges of deep space exploration, traditionally managed by vast teams of human controllers, are now pushing the boundaries of artificial intelligence. AstroForge, a startup with ambitions to mine asteroids, is betting its future on AI, developing an autonomous control system to navigate the unforgiving void.
The Earthbound Challenge of Deep Space
When NASA embarks on missions like Osiris-Rex, which successfully rendezvoused with an asteroid in 2018, the operational model is meticulously structured and resource-intensive. These missions rely on comprehensive redundancy, sophisticated automation, and, critically, large teams of human flight controllers. The Osiris-Rex mission, for instance, employed around 100 operators per eight-hour shift, constantly communicating with the spacecraft across millions of miles.
While this method has proven incredibly successful for government-backed agencies with substantial budgets, it presents an insurmountable barrier for lean startups. The sheer cost and logistical complexity of maintaining such a robust ground infrastructure and human capital are prohibitive for companies like AstroForge, which are funded by venture capital rather than federal coffers. This economic reality forces a paradigm shift: if you can’t afford an army of controllers, you must empower the spacecraft itself to think.
From Anomaly to Autonomy: AstroForge’s Pivotal Shift
Founded in 2022 and backed by $56 million in venture funding, AstroForge quickly encountered the harsh realities of deep space. The company launched two prototype spacecraft, both of which experienced anomalies that severely hampered their mission objectives. A critical lesson emerged from their 2025 Odin mission. Launched into deep space, Odin proved notoriously difficult to communicate with. The limited number of large Earth-based antennas capable of reaching spacecraft hundreds of thousands of miles away, coupled with narrow communication windows, meant AstroForge struggled to establish and maintain control.
Ultimately, AstroForge couldn’t regain control of Odin. This experience wasn’t a failure, but a catalyst. It forced a fundamental question: What if the spacecraft possessed enough intelligence onboard to solve its own problems, independent of Earth’s sporadic communication?
Matthew Gialich, AstroForge’s co-founder and CEO, articulated the strategic dilemma: “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.” He weighed the alternatives starkly: “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?” The answer became clear: AI.
Introducing Solo: The AI Brain for Deep Space
AstroForge’s answer to this challenge is “Solo,” an in-house developed, transformer-based autonomous control stack. Unlike traditional spacecraft autonomy, which predominantly relies on deterministic, rule-based algorithms due to inherent concerns about the unpredictability of neural networks, Solo represents a bold leap forward. While the first use of a neural network for satellite positioning in orbit occurred only last year, AstroForge is pushing the envelope significantly further.
Armand Awad, AstroForge’s head of flight software, explained the hybrid approach. Solo integrates conventional control algorithms—the trusted backbone of spaceflight—with specialized models trained on extensive test data for specific subsystems, such as power generation or navigation. Overseeing this intricate architecture is a general intelligence layer, a transformer model trained on data from approximately 2,500 sensors within the spacecraft. This multi-layered intelligence allows Solo to understand, predict, and react to a vast array of operational conditions.
Gialich emphasized that this isn’t an attempt at generalized artificial intelligence for all scenarios, but rather a focused application: “I’m not saying I’m going to make general spacecraft autonomy or general autonomy for the world,” he clarified. “I’m making a constrained autonomy at a very low sensor input, following the basic training of a transformer model.” This “constrained autonomy” means Solo is designed to excel within the specific, albeit complex, environment of spacecraft operations, particularly anomaly resolution.
Awad envisions Solo performing tasks like realizing it has lost its precise position in space, correlating a power anomaly to potential issues with its star tracker—a device crucial for navigation—and then autonomously resolving the problem. “Probably turning it on and off in this case,” he added, highlighting that sometimes the simplest solution, executed intelligently, is the most effective.
The Road Ahead: Shadow Mode and True Independence
AstroForge is not rushing Solo into a solo mission without rigorous testing. Their third vehicle, DeepSpace-2, is slated for launch by the end of 2026, alongside Intuitive Machines’ third lunar mission. Solo will be onboard DeepSpace-2, operating in “shadow mode.” In this critical phase, Solo will process real-time spacecraft data and make autonomous decisions, but these decisions will not directly control the spacecraft. Instead, AstroForge engineers on Earth will monitor Solo’s performance, comparing its proposed actions with their own commands and observing its learning and anomaly resolution capabilities without flight risk.
The culmination of this development and testing will be the ambitious Autonomy-1 mission, scheduled for 2027 on the first rocket launched by Stoke Space. This mission, which will also gather scientific data about the Sun with NASA backing, aims to push the boundaries of true spacecraft independence. The ultimate goal is to launch Autonomy-1 with minimal, if any, communication capabilities from Earth.
Gialich’s vision for Autonomy-1 is starkly bold: “I don’t plan on flying radios that can receive from Earth on Autonomy-1,” he declared. “We have to go all in right now. The team’s probably going to talk me into it by the time we fly it. But right now, I’m telling them no radios.” This statement underscores not just a technological ambition, but a profound philosophical shift in how humanity interacts with its robotic explorers in space. Such a mission would represent an unprecedented level of trust in artificial intelligence to operate an independent spacecraft, truly freeing it from the tether of Earth.
The Bottom Line
AstroForge’s audacious pursuit of AI-driven spacecraft autonomy marks a pivotal moment in space exploration. By leveraging transformer models to create “Solo,” the company is not just overcoming the cost and logistical constraints that limit startups in deep space; it is fundamentally redefining the relationship between mission control and spacecraft. If successful, AstroForge’s “Autonomy-1” mission will not only validate a new era of self-sufficient space operations but also unlock the potential for more frequent, complex, and economically viable missions, from asteroid mining to interstellar probes, by empowering machines to think for themselves in the vast, silent reaches of the cosmos.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
{content}
Source:{feed_title}