Golden Dome has $185B and real satellites — but the fusion engine still cant get the data it needs

The United States government has agencies with sensors capable of supplementing missile warning. Those agencies are not sharing the raw data from those sensors with the fusion engine that needs it. This is not a secret. It is also, apparently, not a priority anyone has solved.

That was the blunt message from a panel at the Satellite 2026 conference in Washington on March 23, as reported by SpaceNews correspondent Debra Werner. The panelists were Devin Elder, senior business development director at Northrop Grumman Strategic Space Systems; Robin Dickey, director of policy and government affairs at Slingshot Aerospace; and Paul Wloszek, vice president at L3Harris Spectral Solutions. The session was moderated by Clayton Swope, deputy director of the Aerospace Security Project at CSIS.

Title 10 of the U.S. Code governs military roles and missions. Title 50 governs intelligence. The two do not overlap in ways designed for real-time data federation. "Data-control structures and maybe classification reasons" mean organizations "don't necessarily want to share raw data," Elder said. What gets shared instead are "final products that are fully vetted, that they can stand behind" — processed outputs that have already lost the timing resolution and granularity a fusion engine needs.

The physics make this a real problem, not a bureaucratic inconvenience. Missiles move at a minimum of seven kilometers per second. A hypersonic glide vehicle maneuvers during flight. The targeting data that informs an intercept has, as Elder put it, "a finite lifetime." For a 3D fusion engine to take radar, optical, and infrared inputs from multiple sensors and output a targeting solution, it must do so in seconds. Pre-processed final products do not make that deadline.

The hardware pipeline is real and funded. The Pentagon is spending $185 billion on Golden Dome, the layered homeland missile defense architecture. The Space Development Agency has awarded $3.5 billion to L3Harris Technologies, Lockheed Martin, Northrop Grumman, and Rocket Lab for 72 Tranche 3 tracking satellites. Those satellites carry onboard AI processors capable of picking out "the literal hypersonic needle in the haystack of infrared signatures," as Wloszek put it. Numerica Corporation is providing onboard data fusion software for Tranche 1 satellites, enabling edge processing without a ground relay. Two HBTSS demonstration satellites launched in 2024 proved handoff between sensors and an interceptor.

A January 2026 GAO report found that SDA overestimated technical readiness, carried unplanned contractor work, and had no architecture-level schedule or reliable cost estimate for the Proliferated Warfighter Space Architecture — projected at $35 billion through fiscal year 2029 for 300 to 500 satellites. The GAO issued six recommendations. DoD concurred with five. The hardware gaps are manageable if the money keeps flowing, but the money is only part of the problem.

But the pipeline only works if it gets the data. Dickey described the multi-phenomenology problem directly: "There's no one sensor to rule them all. Looking up, you're going to see very different aspects of an object than you would looking down. The background is different. The object itself could look different." Radar, infrared, optical — each adds information the others miss. "There are agencies that have sensors that can provide data that would be very supplemental to the missile-defense mission," Elder said, "but they designed it for a completely different purpose." Those sensors are operated by agencies under Title 50 authorities. Their data does not flow to the Title 10 military fusion engine in raw form. It flows as finished intelligence products.

Wloszek described what the integrated system requires: "For a 3D fusion engine to take all that information and output an answer in seconds... we all have to speak the same language." What the hardware has replaced — literally humans in the loop interpreting thermal signatures and deciding if something is a threat — was a real engineering achievement. The chain still breaks at the agency boundary.

None of this is fixed by building more satellites. Northrop holds contracts for 150 satellites across the first three tranches of the SDA constellation. Rocket Lab's first SDA tracking contract uses its Lightning bus and Phoenix infrared sensor.

General Michael Guetlein, the Golden Dome program lead, told the McAleese Defense Programs conference this month that the technical challenge is not the hard part — scaling fast and affordably is. The first operational demo is required by summer 2028. Full operational architecture: 2035. Based on public statements and filings reviewed by type0, no single agency has been designated to own the resolution of the Title 10/50 data-sharing barrier that prevents raw sensor data from those agencies from reaching the fusion engine.

The hardware is funded. The data stovepipe is a legal fact.