A strange thing happened in July when the U.S. Navy’s MUOS-5 satellite was on its way to geosynchronous orbit. The thruster it was using to raise its orbit stopped working unexpectedly. It’s never good when a part fails like this. It’s especially bad when this happens to a military satellite worth $475 million including launch costs, leaving it in the wrong orbit — at least temporarily. It turns out that the thruster in question — a BT-4 apogee engine made by IHI Aerospace of Japan — may have components in common with the engine on the Intelsat 33e satellite and NASA’s Juno probe, both of which recently experienced propulsion malfunctions.

If the problem for the military was limited to the MUOS-5 satellite, the impact would be minimal. The Mobile User Objective System (MUOS) provides mobile voice and data links to connect commanders with troops in the field, and MUOS-5 was the last planned satellite in the constellation — intended to be an on-orbit spare. The other four MUOS satellites already made it to orbit without problem, so life would have gone on even if the Navy had not been able to recover this satellite.

Unfortunately, the problem does not stop there. Three of the U.S. military’s major satellite programs — AEHF, SBIRS, and MUOS — use components common to the BT-4 apogee engine. Advanced Extremely High Frequency (AEHF) satellites provide secure, jam-resistant communications for nuclear command-and-control and tactical users. The Space-Based Infrared System (SBIRS) is used to detect missile launches, and to provide critical battlespace intelligence for U.S. and allied forces around the world.

The first AEHF satellite experienced a problem with its BT-4 apogee engine when it was launched in 2010, but the Air Force was ultimately able to move the satellite to its intended orbit using other thrusters. Engineers determined that this failure was due to a cleaning cloth left in a propellant line, which clogged the engine. A SBIRS GEO satellite, the third in the constellation, was planned for launch in October, but the Air Force delayed its launch pending an investigation into the BT-4 issue.

The BT-4 engine issue should serve as a wake-up call for senior leaders in the Pentagon and Congress. It highlights the fact that critical U.S. military space systems are dependent on a small number of large, expensive satellites, and that many of these satellites use similar parts, such as the BT-4 apogee engine, on-board processors, and power components. If a critical flaw is discovered on one of these components in the future, it could have widespread effects across multiple military-space systems simultaneously. Moreover, the fact that these constellations rely on a small number of satellites that each take years to build (and, therefore, years to replace) makes each satellite a juicy target for adversaries. The potential benefits an adversary could gain from disabling or destroying just one or two of these satellites are substantial.

The current architecture for military-space systems places too many eggs in too few baskets. The solution is to stop thinking in terms of constellations of satellites and to start thinking in terms constellations of payloads. Doug Loverro, the current deputy assistant secretary of defense for space policy, has advocated for a payload-centric approach for years. A more recent champion of this approach more generally in defense acquisitions is Jamie Morin, the director of the Defense Department’s Cost Assessment and Program Evaluation. In a tongue-in-cheek interview with VICE News, Morin made the case for a “payloads not platforms” approach by making comparisons to the Star Wars-vintage Death Star, which he notes is “one platform [that] takes a huge amount of infrastructure and is really only good for one mission.” While current military-satellite constellations aren’t intended to destroy entire planets — something that would be contrary to stated administration policy — they are similar to the Death Star in that they vest large amounts of resources in a small number of vulnerable platforms.

Rather than continuing a Death Star-like approach, the U.S. military needs to rethink its space architecture and next-generation space systems. The timing is ideal for such a shift because five of the six major unclassified military space programs are completed or nearing completion (AEHF, MUOS, SBIRS, Wideband Global Satcom, and the Defense Meteorological Satellite Program), and follow-on systems are needed for each of these. The start of a new administration and the release of the FY 2018 budget request in a few months provides the perfect opportunity to begin charting a new direction.

Applying the “payloads not platforms” approach to military-space systems would mean designing future space architectures that disperse payloads over a large number of diverse satellite platforms in a variety of orbits. With a more dispersed and diverse architecture, adversaries would need to attack a larger number of targets spread over a variety of platforms — which could include U.S. military, commercial, and allied nation satellites — to achieve the same effects as attacking the small number of juicy targets presented to them today. Where it is strategically advantageous, the U.S. may also want to disaggregate certain capabilities (i.e. not host certain payloads together on the same satellite) to reduce potential ambiguity if these systems are attacked. Launching more payloads on a more frequent basis provides more opportunities to introduce upgrades and new capabilities as they become available. And more importantly, distributing payloads across a more diverse set of platforms would limit the impact of something like the BT-4 apogee-engine problem in the future.

As the clock ticks down to the end of this administration, there is important work that remains to be done. The program review process currently underway in the Pentagon for the FY 2018 budget request is this administration’s best and final chance to move military-space programs to a higher, more sustainable orbit.

Todd Harrison is the director of the Aerospace Security Project and the director of defense budget analysis at CSIS