Satellites Freeze at Launch. Ground Systems Don't

Golden Dome has a sensor problem. Not a hardware problem — a staying-current problem.

That is the argument Orion Space Solutions, a subsidiary of defense contractor Arcfield acquired in November 2023, laid out in a SpaceNews op-ed last week. The authors — Junk Wilson, senior vice president of sensors and models, and John Noto, chief scientist — argue that ground-based electro-optical and infrared sensors should form the backbone of Golden Domes sensor architecture, not satellites. Their core claim: a satellite is a time capsule launched with hardware frozen at the design stage, while ground-based optics can be continuously modernized. They call it upgrade-at-threat-speed. That phrase is Orion intellectual property, not industry standard, but the underlying physics does not care who coined it.

Satellites cannot be upgraded post-launch. Ground-based systems can be. That is a real constraint, not marketing copy. The United States currently operates 44 Ground-Based Midcourse Defense interceptors in Alaska and California, and they cannot reliably distinguish warheads from decoys — a limitation that predates Golden Dome entirely. Adding sensors that see better and think faster addresses part of that gap. Ground-based systems are also passive: they do not emit signals of their own, making them harder to locate and jam than space-based radar assets. Orion makes this argument, and it is technically sound.

But Orion leaves two questions unanswered. Ground-based sensors can be degraded by weather — relevant when tracking hypersonic threats traveling at the low end of 16,000 miles per hour. More significantly, fixed ground sites are vulnerable to kinetic attack. Orion does not address either limitation. That selective framing is worth noting when evaluating the pitch.

The harder problem is not the hardware.

Gen. Michael Guetlein, Golden Domes program director, has called the command-and-control system the programs secret sauce. He is not wrong. The real integration challenge for Golden Dome is not a technology gap — it is an organizational one. Multiple agencies operate sensors designed for entirely different primary missions: MDA, Space Force, Army, Navy, the National Reconnaissance Office, and the intelligence community. They have different legal authorities, different classification levels, and different institutional incentives to share data in real time. An industry panel discussed this in March: I do not think it is as much a technical challenge as it is a political or maybe bureaucratic challenge. There are agencies that have sensors which can provide data that would be very useful for the missile defense mission, but they are designed for a completely different purpose.

That is the part no sensor vendor can solve. You could spec the best ground-based optics array on the planet, and if the data cannot cross classification boundaries to the interceptor network in time, the array does not matter.

Guetlein is building an accountability mechanism around this. A nine-company consortium briefs him every Thursday. The group operates as a unit — they decide what they are building, who builds each component, and then hold themselves accountable on a weekly basis. Underperformers can be voted off. That is an unusual governance structure, and it exists precisely because the organizational problem is harder than the engineering.

The HBTSS tracker is mostly working.

The Hypersonic and Ballistic Tracking Space Sensor program is the most mature space-based sensor effort in the architecture. Two demonstration satellites launched in 2024. One unit, built by L3Harris, is meeting MDA requirements and has demonstrated the ability to track missile threats and pass targeting data to interceptors. The second unit, built by Northrop Grumman, did not meet requirements. One of two is working. Hardware is hard. Boeing opened a 9,000 square foot EO/IR production line in El Segundo, California in February, marking a transition from rapid prototyping to rate production. That is real industrial momentum.

The cost problem has not gone away.

Golden Dome is currently budgeted at $185 billion, up from $175 billion, with an additional $10 billion approved specifically for space capabilities including HBTSS and the Space Data Network. According to Reuters, the Congressional Budget Office issued a May 2025 letter estimating that a space-based interceptor constellation could cost between $161 billion and $542 billion over 20 years, after accounting for modern launch costs roughly 90 percent lower than the assumptions used in prior estimates from 2004 and 2012. The same analysis showed that under older cost assumptions — before the launch cost adjustment — those figures would have been $264 billion and $831 billion respectively. That $831 billion figure appears in some sources as the CBO estimate; it is the same analysis as the $161-542 billion range, before the launch cost recalculation, and both numbers come from the same CBO document. The American Enterprise Institute estimate — which Guetlein has publicly rejected — comes in at $3.6 trillion. The spread is not a rounding error. It reflects fundamental disagreement about what the system is, how layered the defense needs to be, and whether space-based interceptors are in the architecture at all. Guetlein identified space-based interceptors as the programs highest-risk element, citing scalability and affordability as the central challenge. He is probably right on both counts.

The United States must defend a territory roughly 16 times larger than Israels — 13,699 miles of border versus 834 miles. Scale is not a detail.

The deadline is the story.

Guetlein is required to demonstrate operational capability by summer 2028. That is roughly two years away. The sensor question — space versus ground, which modality, whose data feeds which interceptor — is genuinely open. Orion Space Solutions argument that ground-based optics are the more upgradeable backbone is technically plausible and worth taking seriously on its merits, not dismissing because the authors work for a company that would benefit from the contract. But the harder truth is that no sensor architecture matters if the data fusion problem — bureaucratic and legal, not technical — remains unsolved by 2028.

Guetlein told the McAleese Defense Programs conference that the biggest challenge is not technology but scaling solutions fast enough and affordably to be effective against the threat. That is the most honest thing anyone in the program has said publicly. Everything else is a vendor pitch.