Inthe mid-20th century, the world’s superpowers raced to the moon. In the 21st, they’re racing for control of the orbital algorithms that may determine who wins — or survives — the wars of the future.
From Beijing to Washington to Moscow, artificial intelligence (AI) is propelling military power on orbit. The United States still commands the largest constellation of military satellites, but the Chinese Communist Party (CCP) is moving “breathtakingly fast,” according to U.S. Space Command chief Gen. Stephen N. Whiting. Beijing’s new Three-Body Computing Constellation, a network of supercomputing satellites, will process data on orbit using AI models — effectively turning space into a massive cloud network.
For Washington, this is no longer just about rockets or radar. It’s about cognition. The U.S. Space Force (USSF) — the newest and most digitally aware service in the U.S. — has made AI the centerpiece of its 2025 Strategic Action Plan, declaring that “superiority will be defined by our ability to integrate with allies and advance real-time analytics and emerging AI technologies to outpace adversaries.”
In the eyes of the Space Force, the race has become a contest of code, but other branches are not far behind. At a January 2025 defense summit in Virginia, scientists from the Naval Research Laboratory unveiled a prototype called Autosat — a fully autonomous satellite that can calibrate, navigate and analyze data without human guidance. “We’ve proven out the principles and are looking for the next step,” said Steven Meier, the lab’s director of space technology, on the news site Defense One. “We want to get funding to actually build a system and launch it.”
Autosat’s intelligence grows with every image it sees — airports, roads, buildings — letting it independently recognize objects and patterns. In the next decade, Meier predicts a network of such AI-driven satellites capable of operating without GPS or ground control.
The implications are profound. An autonomous satellite fleet could continue functioning even if cyberattacks or jamming sever communication with Earth. “Advances in autonomous, on-orbit decision-making hold the potential to significantly reduce response times and strengthen the effectiveness of defensive space operations,” Krista Langeland of the Rand Space Enterprise Initiative told Apogee. “By diminishing reliance on the SCN [Satellite Control Network], greater autonomy could lessen potential vulnerabilities there and would circumvent any capacity limitations and lack of antenna availability.”
In plain English, these machines would think — and react — faster than any human operator, adjusting even for ground station attacks.
As space situational awareness sensors proliferate, more data explaining orbital activity will become available in real time. “A machine can be trained to sift through that data at speeds way faster than anything possible using a human alone,” Clayton Swope, deputy director of the Aerospace Security Project at the Center for Strategic and International Studies (CSIS), told Apogee. “This might help spot possible counterspace activities before or as they are unfolding, giving satellites a chance to mount a successful defense.”
No competitor looms larger than the CCP. According to the U.S.-China Economic and Security Review Commission, Beijing has tripled its satellites in low Earth orbit since 2018, doubled those in geostationary orbit and integrated them into a sweeping system for surveillance, targeting and electronic warfare.
These advances are part of what a 2018 Rand Corp. report identified as “systems confrontation” and “systems destruction warfare” — a CCP doctrine that prioritizes disabling an adversary’s communications and data flow. In a conflict, Chinese hackers would aim to take down networks controlling U.S. satellites, leaving them blind or unable to transmit.
That’s where AI could become the Pentagon’s shield. A new generation of self-healing, self-learning satellites could detect and counter such intrusions autonomously, bypassing jammed ground channels. Unlike Earth-based detection systems, AI offers a scalable way to monitor thousands of nodes simultaneously, in effect turning U.S. satellites into sentient firewalls. The U.S. Navy’s AI-driven space systems, Defense News reported in January 2025, already are designed to “navigate without GPS, communications, and ground control.”
For the CCP, however, AI is not just a defensive tool — it’s a platform for dominance. The Three-Body Computing Constellation, launched in the summer of 2025 with provincial government backing by Hangzhou-based startup ADA Space, includes 12 satellites equipped with onboard AI processors. Each acts as an independent data center.
an AI-powered space computing satellite constellation into space. XINHUA NEWS
“China’s approach closely parallels its electric-vehicle strategy — leapfrogging older technologies to lead in next-gen domains,” space researcher Sylwia Gorska of the University of Lancashire (England) told Bloomberg in July 2025. By processing data on orbit, CCP satellites reduce latency and vulnerability to terrestrial attacks while establishing a digital infrastructure less accessible to Western surveillance.
“This marks a major shift in the U.S.-China space competition,” Arizona State University faculty associate Namrata Goswami told Bloomberg. “China is creating asymmetric advantages — autonomous surveillance, environmental monitoring, and military applications less vulnerable to Earth-based attack.”
If the Cold War was fought over nuclear stockpiles, the United States’ next great rivalry is being waged with neural networks. At the Pentagon’s youngest branch, the AI revolution has become institutional doctrine. The Space Force’s 40-page Data & Artificial Intelligence FY 2025 Strategic Action Plan lays out an agenda that could have been drafted in Silicon Valley. Its four Lines of Effort read like a tech startup mission statement:
- Mature enterprisewide AI governance
- Advance a data-driven culture
- Rapidly adopt advanced analytics
- Strengthen government-industry-international partnerships
In practice, this means turning the force’s Guardians into coders and data scientists. Professional military education courses soon will include AI literacy modules. New AI boot camps will train Guardians to use and develop machine learning (ML) models. The plan also calls for Wisdom, a recurring speaker series bringing together tech leaders from the Pentagon, academia and industry, and Momentum, a quarterly publication showcasing applied AI research.
Behind the bureaucratic vernacular lies a simple mission: teach a new generation of warfighters to think — and fight — like digital natives with the best tools at their disposal. “Data and AI are critical for a warfighting service that is purpose-built for space superiority,” Space Force Col. Nathan L. Iven, special assistant to the vice chief of space operations, said in a March 2025 announcement about the force’s AI action plan.
While the Pentagon builds policy, Silicon Valley builds technology. During a Mitchell Institute for Aerospace Studies podcast in September 2025, Primer Technologies CEO Sean Moriarty described how his company — originally incubated by In-Q-Tel, the CIA’s venture arm — helps warfighters navigate the flood of unstructured battlefield data. Primer’s software uses large language models to parse intelligence reports, social media and field logs in real time.
“We want people to start their day where they used to end it,” Moriarty told host Heather “Lucky” Penney. “Our goal is to do the heavy lifting that isn’t strategic so analysts can bring their creativity and imagination to bear from the beginning.”
That ethos — AI as a teammate, not a replacement — echoes across the defense tech world. Primer’s tools help commanders sift through mountains of drone-related data in the counter-UAS (unmanned aerial systems) mission, fusing open-source information with classified feeds to track evolving threats in Ukraine, Israel and elsewhere.
By automating discovery and reducing human triage, Moriarty said, “we shift the burden from data collection to intelligence.” In one case, the system reportedly helped a U.S. combatant command detect rebel movements threatening aid workers — a process that once took 50 days was completed in mere hours. “That’s practical AI,” Moriarty said. “Software that works.”
Trusting the Machine
Yet as machines grow smarter, some question how much they can be trusted. Space Force officials acknowledge that fully autonomous decision-making — especially in systems capable of lethal or strategic actions — demands caution. “What trust level is required where I can have one Guardian reviewing 10, 20 or 1,000 space objects with automated help? I don’t know what that number is. It’s the conversation that we’re having,” Seth Whitworth, Space Force acting associate deputy chief of space operations for cyber and data, said at the February 2025 Space+AI Summit.
The Space Force is experimenting with human-in-the-loop oversight, where AI conducts the bulk of monitoring while humans retain veto power. But as systems scale up, human attention itself may become a liability. “A tired operator presented with too much data could actually detract from safety,” noted Pat Biltgen of Booz Allen Hamilton, a Virginia-based company that builds advanced technology for government.
AI supremacy in space isn’t just about speed and autonomy — it’s about survival in the face of invisible enemies. A leaked CIA report circulated by media in 2023 asserted that CCP scientists were developing advanced cyber weapons to seize control of enemy satellites during conflicts. An International Business Times article in April of that year said the weapons could render satellites useless for data signals and surveillance during wartime. This was being done as part of the CCP’s core strategy to “seize control of a satellite, rendering it ineffective to support communications, weapons, or intelligence, surveillance, and reconnaissance systems.”
U.S. satellites have since been progressively hardened with AI-enabled cybersecurity. The systems use ML to detect anomalies in telemetry data, predict intrusions and quarantine malicious code before it spreads. One such tool the Space Force hopes to add to its collection would further detect cyberattacks by monitoring satellites’ behavior and telemetry outputs. Cyber Resilience On-Orbit (CROO), the technology by Albuquerque, New Mexico, startup Proof Labs, uses ML/AI to identify anomalous behavior in space. The Space Force anticipates CROO’s delivery, including potential hardware and software options, in 2026.
Faster acquisition?
For many years, defense innovation followed a predictable pattern: The Pentagon outlined specific needs, contractors developed the requested systems and political leaders acquired the equipment within a budget cycle. AI has upended that model. Now, commercial startups are outpacing military R&D, forcing the government to adapt to the tempo of the tech industry.
“The speed of innovation is no longer set by defense bureaucracy,” host Penney said during the Mitchell podcast. “It’s set by the market.” Companies like Primer, Anduril and Palantir already are fielding dual-use AI platforms that blur the line between civilian and military. The Space Force’s plan explicitly calls for deeper collaboration with such firms, emphasizing commercial data integration and shared AI literacy.
“The United States has an extraordinary opportunity to evolve its capabilities without breaking the bank,” Primer CEO Moriarty said. “These technologies are already funded by billions in private investment. The key is for government to be a smart consumer.”
In other words, the next generation of deterrence may come not from classified labs but from venture capital. As of 2025, more than 11,000 active satellites orbit Earth — triple the number of just a decade ago. With commercial megaconstellations like SpaceX’s Starlink and Amazon’s Project Kuiper crowding the skies, space domain awareness is a monumental challenge.
The Space Force’s Unified Data Library aggregates information from military sensors, private satellites and international partners to maintain a shared picture of orbital traffic. Integrating that data requires sophisticated ML models capable of sifting petabytes of telemetry data to detect anomalies.
“Long gone are the days when one airman could monitor one satellite,” the Space Force’s Whitworth said. “The question now is how many space objects one Guardian can track with AI assistance?”
The answer: potentially thousands. To manage that scale, the Space Force has begun benchmarking large language models for mission-critical tasks. The aim is to develop a chatlike capability that allows Guardians to query orbital data conversationally — essentially turning an AI app like ChatGPT into a space operations assistant. It’s a glimpse of a future where the command console looks less like the 1983 film “War Games” and more like Microsoft Copilot. Despite the automation buildup, most defense leaders insist AI will augment, not replace, human judgment. “AI is there to enhance human performance,” Penney said. “You’re not replacing people — you’re making them better.”
In practice, that means software handles the drudgery — data cleaning, pattern recognition, anomaly flagging — while humans focus on strategy.
Still, building trust between humans and algorithms remains a major hurdle. The Space Force is investing heavily in AI literacy precisely because trust must be earned, not assumed. A Guardian who doesn’t understand how a machine reached its conclusion will hesitate to act on it — and hesitation on orbit can mean the difference between collision and avoidance.
“The decision to initiate a defensive maneuver often requires a depth of situational awareness that exceeds what an autonomous single satellite may have available to it,” Langeland said. “The development of these autonomous capabilities should therefore begin with a clear assessment of the data and contextual inputs required for effective decision-making and defense.”
For all its promise, AI carries inherent dangers. Faulty data can corrupt models. Adversaries can feed disinformation into open-source datasets. And as machines gain autonomy, accountability becomes murky. “Information quality matters an awful lot,” Moriarty cautioned. “You have to know the provenance of your data.”
When Russia began its invasion of Ukraine in 2022, Western analysts confronted this problem — a torrent of contradictory online claims about troop movements. Primer’s system responded by surfacing “contested claims” and flagging inconsistencies for human review — illustrating both the power and peril of automated intelligence.
“You certainly don’t want to underestimate an adversary,” Moriarty said. “But overestimating one isn’t without consequence either.” In an era when misinformation travels at light speed, even perfect code can be poisoned by bad input.
The Pentagon’s emerging strategy could be described as deterrence through cognition. If nuclear deterrence relied on mutually assured destruction, AI deterrence may rely on mutually assured detection — an environment where every move is observed, analyzed and anticipated in real time.
Autonomous satellites soon could identify hostile launches or cyber intrusions the instant they occur, triggering countermeasures before human commanders even wake up. Such responsiveness could prevent escalation — or accelerate it uncontrollably. That tension — between control and chaos, autonomy and oversight — is the central drama of the AI space race.
“Decision-making machines need rules of the road just as much as people,” CSIS’ Swope said. “Without agreed-to rules of the road, whether satellites are controlled by humans or AI-enabled algorithms, it is hard to see significant breakthroughs in space safety.”
Digital Warriors
Inside the Space Force’s data centers, a new generation of digital warriors is taking shape. They are not pilots or astronauts, but coders — Guardians who defend constellations instead of continents. Their weapons are neural networks, their battlefield, the thermosphere. They are trained to think algorithmically, act collaboratively and trust machines as teammates. “The USSF recognized the need for digital fluency within the Guardian workforce, and based on the latest national guidance, the Space Force needs to emphasize AI literacy to employ these tools to solve operational challenges,” Iven said.
It’s likely that fluency will evolve and sharpen. Former Space Force acquisition chief Frank Calvelli predicted that by 2035 most U.S. military satellites will be “significantly more autonomous.” The era of joystick-controlled spacecraft will give way to fleets that navigate, analyze and even negotiate orbital paths on their own. “I could really envision 10 years, 15 years down the road, satellites being significantly more autonomous than they are today, like not needing operators, like being able to respond and take actions and do things on their own,” Calvelli said during the August 2024 National Defense Industrial Association Emerging Technologies for Defense conference, according to Defense One.
The push toward autonomy in space makes sense, he said, given where the weak link in satellite systems lies, at ground-based control stations that are more vulnerable to cyberattacks. “I believe that from a threat perspective, ground antennas and ground stations are, to me, something that’s at risk in a conflict, so I think down the road, you want to take advantage of where the industry is going and really make your constellations as autonomous as possible and get human control out of the loop,” Calvelli said.
Meanwhile, the CCP’s plan for 2,800 AI-enabled satellites could give it a global surveillance grid powered by in-space computing. Russia, though economically constrained, continues to refine its anti-satellite and electronic warfare systems. The result is an increasingly contested, congested and competitive domain — a digital Wild West at a low Earth orbit speed of over 27,000 kph.
Why does it matter who dominates space AI? Because every other domain — air, sea, land and cyber — depends on it. GPS, communications, missile warning and navigation all rely on orbital systems. Lose space and you lose the war before it begins. “The winner will control the vital infrastructure of modern warfare,” Melanie Garson, an associate professor in International Conflict Resolution & International Security at University College London, told Business Insider. “It’s not just about satellites — it’s about who owns the data highways of the planet.”