Shield Space plans autonomous satellite-defense swarm tests

A cubesat unit, or U, is a cube measuring 10 centimeters, about 4 inches, on a side. The 8U and 3U names express how many such unit volumes each craft contains.

NASA’s civilian Starling project, launched in 2023, demonstrated that four small satellites could cooperate and operate autonomously during a science mission. The spacecraft relied little on real-time commands from ground control, but Starling did not test combat behavior.

No leading space power has publicly demonstrated swarm combat, though major powers may be reluctant to disclose such capabilities. Shield Space’s 2027 test would address a more demanding application than Starling.

Can the hardware fit inside a cubesat?

Ritchie called size, weight and power the central engineering problem and said the team has only roughly 100 by 100 centimeters to work with. Some mission elements have required new designs from scratch. The aim is to save space or use integrated systems while fitting propulsion, power systems, sensors, AI computing, laser communications and weapons systems within the weight allowance.

Shield Space has commissioned EnduroSat to build the autonomous 8U cubesat, including chemical propulsion and its own imagery-collecting capability. Shield Space will install its software and use the craft for the RPO demonstration.

“The good thing about cubesats is they’re very standardized and they’re very off the shelf,” said Dan Molland, co-founder and chief technology officer. He said cubesats still need some adjustment, but they are more plug-and-play than conventional platforms, which are usually custom-built. Shield Space sees that standardization as a route to cheaper systems than bespoke designs.

Can artificial intelligence control the swarm?

Ritchie said autonomous models need an extraordinary amount of training data, but data comparable to that available from the Ukraine war are not readily accessible for space combat. Synthetic data can help train the system, but he said it is not quite good enough for space.

Molland said Starling mainly asks a guidance system to stay within a specified range of another satellite. Controlling a military swarm raises a more complex set of questions, and its AI must follow rules of engagement, which govern when force may be used.

Both believe the technical hurdles can be overcome, though Ritchie says missions such as Broadsword must generate the missing data. Their schedule includes several sharply different steps between a two-craft maneuver and a distributed defensive network.

What does history say about the idea?

Bleddyn Bowen, an associate professor of astropolitics at Durham University and author of War in Space: Strategy, Spacepower, Geopolitics (Edinburgh University Press, 2020), compared Shield Space’s concept with Brilliant Pebbles. That late-Cold War research program within the Strategic Defense Initiative sought to defend the United States from Soviet intercontinental ballistic missiles with thousands of small autonomous orbiting interceptors that would destroy missiles by collision.

Bowen said enough investment makes a lot feasible, but warned that any such system would require many years of development and testing. The unresolved commercial question, he said, is finding someone to buy it.

What milestones come after the 2027 demonstration?

Shield Space hopes to move quickly through four levels of hardware and scale:

• Late 2027: Project Broadsword’s two-craft RPO demonstration.
• Next: tests involving two full-size 16U cubesats.
• Then: one carrier carrying 10 such cubesats.
• Later: a “super carrier” containing 10 carriers, deployed across multiple orbits.

Only the first milestone has an announced date. Ritchie calls the long-term concept “space-based carrier strike.” In that scenario, responsive launches or pre-positioned carriers could protect entire orbits, but Shield Space has not demonstrated that capability.

An editor’s note, updated at 10:45 a.m. ET on Nov. 5, says the original story suggested Shield Space might be developing kinetic-impact, or collision-based, satellite defense. The company says it is not doing so.

For now, Project Broadsword is the next test. The proposed 100-cubesat fleet remains a long-term vision rather than an operational capability.