National Dome? Listen to the Americans

WASHINGTON — To launch the Golden Dome program without it exploding in costs, the U.S. Department of Defense needs a breakthrough in acquisitions and artificial intelligence. The crux, in simple terms, is one: “it costs too much”.

Speaking at the Miami Space Summit, Marcia Holmes, deputy director of the DoD for Golden Dome, stated that the “real” challenge is not just bringing together new technologies, but integrating space assets and weapon systems into a single architecture capable of providing a layered defense, on a large scale, and with sustainable costs.

The problem: integrate everything (and do it without overspending)

According to Holmes, the goal is to effectively connect:

• missile tracking satellites,
• potential space-based interceptors (SBI),
• a comprehensive command and control (C2) network that connects sensors and “shooters”.

The critical point is to make this architecture operational quickly but also economically manageable. For this reason, she explained, the Secretary of Defense would have initiated a “broad and radical” acquisition transformation strategy, with the idea of speeding up development and deployment while reducing waste and time.

AI and autonomy: less personnel, lower costs, more speed

Holmes pointed to AI and autonomy as fundamental accelerators:

• More autonomy = less personnel needed = lower sustainment costs.
• More AI = ability to analyze enormous data flows and quickly present options to decision-makers to choose the most suitable response.

In particular, AI would be decisive for an AI-powered C2 capable of transferring data almost in real-time from sensors to interceptors and other defense systems on land, sea, and air. “C2 is crucial for the success of the Golden Dome”, she reiterated.

SBI: the great cost multiplier

One of the most contested points remains the space-based interceptors component: analysts and legislators fear that too many interceptors in orbit are needed to stop multiple missiles launched almost simultaneously.

Here a physical limit comes into play: in low Earth orbit (LEO) an interceptor remains “in useful position” relative to a ground target for only 7–10 minutes. Result: to truly cover the globe and ensure engagements, numerous constellations are needed. The problem is known as the “absentee ratio” and is even tougher in the boost-phase (right after launch), where the window to strike lasts few minutes.

Missile tracking: military space + commercial solutions

Holmes said that a “key element” will be the ability to develop and integrate data from tracking satellites made by both the DoD and the private sector, leveraging existing systems and also focusing on commercial solutions.

In the overall picture:

• the Space Development Agency (SDA) is working on a Tracking Layer in LEO to track ballistic and hypersonic missiles, with global coverage expected by 2029, in addition to “fire control” sensors;
• the program has received remarks from the GAO on risks of technological development, delays, and cost overruns;
• Space Systems Command is also pursuing tracking in medium Earth orbit (MEO): in June 2025 a contract worth 1.2 billion dollars was awarded to BAE Systems for 10 Epoch 2 satellites, with launches expected from 2029.

DIU pushes on commercial: “high fidelity” sensors and tight deadlines

The latest acceleration comes from the Defense Innovation Unit (DIU), which seeks high-precision commercial sensors to:

• detect, track, and discriminate (separate real warheads from decoys/countermeasures),
• enable fire control with highly accurate data for engagements with kinetic kill vehicle (KKV).

The timeline is tight: laboratory demonstration in 6–8 months from assignment and in-orbit testing in 12–24 months.

Final line

For the Pentagon, Golden Dome is played on a balance: layered and integrated defense, but above all less “it costs too much”. And the recipe, according to Holmes, involves two levers: reforming how systems are purchased and developed and using AI/autonomy to make the machine (sensors–C2–interceptors) work faster and with lower costs.