Gateway Is Dead. The Moon Base Killed It.

NASA is about to spend $30 billion on a moon base. The real news is what it killed to get there.

Space Reactor-1, a 20kW fission spacecraft launching to Mars in 2028, is the harder bet. Gateway is the confession.

NASA announced a $20 billion moon base at its Ignition event this week. Buried in the same briefing, the agency confirmed what many in the space community had already assumed: the Gateway, a human-tended orbital outpost around the Moon that has consumed $4.5 billion in development since 2019, is effectively dead. The pause is not being called a cancellation — Congress wrote Gateway into law, and calling it a pause is the only legally clean exit — but NASA says it will repurpose applicable equipment, and the fate of the modules already built is not yet settled. Jared Isaacman said at the Ignition event the Gateway pause should not surprise anyone. That is the story.

The moon base itself will cost at least $30 billion across three phases: $10 billion through 2028 for up to 30 Commercial Lunar Payload Services (CLPS) landings plus one crewed touchdown, another $10 billion for semi-permanent habitation starting in 2029, and a minimum $10 billion more through 2036 for in-situ resource utilization (ISRU) and scaled-up cargo delivery. That $20 billion figure the announcement led with appears in more than one briefing at the same event, attached to different programs and different congressional line items. Isaacman's moon base team cited $20 billion over seven years. Ars Technica reported Isaacman separately invoked $20 billion in failed nuclear programs as context for why SR-1 needed to succeed. The Planetary Society noted the agency expects to spend approximately $6 billion on expanded CLPS operations over the coming decade — separate from the moon base budget. Three separate figures, three different program offices, no single budget line connecting them.

The more technically interesting hardware is Space Reactor-1 (SR-1) Freedom, a nuclear-electric propulsion spacecraft scheduled to launch toward Mars before the end of 2028. Unlike solar-electric systems, nuclear-electric propulsion uses conventional plasma thrusters — the kind that have flown on Dawn and numerous commercial satellites — but substitutes a fission reactor for solar panels to energize the xenon fuel. SR-1 will produce roughly 20 kilowatts from its fission reactor, a fraction of what NASA's earlier Project Prometheus aimed for, but 20 times more electrical power than the radioisotope generators currently operating on Mars rovers and the Voyager probes. Steve Sinacore, NASA's program executive for space reactors, described the decades-long gap not as a technology problem but an execution problem. "Honestly, we have not won the right to be able to do that after $20 billion worth of failed programs over time," Jared Isaacman, NASA administrator, told Ars Technica in reference to the nuclear propulsion track record.

Mission design for SR-1 Freedom will be finalized by June 2026, with large-scale assembly beginning in early 2028. The full vehicle is scheduled to arrive at the launch site by October 2028, ahead of a December launch window. Miss that window, and the next Earth-Mars alignment does not occur until early 2031. The spacecraft will deploy three Ingenuity-derived Skyfall helicopters after hypersonic Mars entry — mid-air release, no landing gear, first-ever deployment of an aircraft from a planetary entry vehicle. Whether any of this happens on schedule depends on whether $20 billion in new nuclear hardware actually gets built and tested in 26 months. The last U.S. nuclear reactor to fly in space was SNAP-10A, launched in 1965.

The hardware already built for Gateway has multiple possible destinations. Ars Technica reports the Power and Propulsion Element (PPE), currently under construction at Lanteris Space Systems in Palo Alto, California, is being cannibalized for SR-1 Freedom — Mars hardware, not surface architecture. The Habitation and Logistics Outpost (HALO), which recently arrived in the United States from Italy for final outfitting, may be repurposed for the lunar base or other programs, per SpaceNews. NASA says it will repurpose applicable equipment; what lands on the surface and what flies to Mars may not be the same thing. Both modules were designed for lunar orbit, not surface operations. Reflying either of them on a surface architecture is not a trivial modification — it is an admission that the orbital station those modules were built for was the wrong architecture. NASA ISS program manager Dana Weigel explained the agency faces no compelling in-orbit manufacturing demand, no substantial tourism market, and no international partners willing to fund long-duration orbital access on a U.S. station.

The geopolitical pressure is real: China has stated it aims to bring humans to the lunar surface by 2030. NASA has since pushed back its own crewed lunar landing to 2028, and the agency is seeking no fewer than two independent commercial launch providers capable of landing crew on the Moon on a six-month cadence. "The clock is running in this great-power competition, and success or failure will be measured in months, not years," Isaacman said at the Ignition announcement. Whether $30 billion across three phases materializes depends on a budget process that has killed more ambitious space architectures than this one.

The nuclear spacecraft is the harder engineering bet. The moon base is the politically survivable one — it spends money on the ground, builds things that land, and distributes contracts across a dozen states. SR-1 launches a fission reactor on a spacecraft that has never flown, toward a planet whose launch window closes in December 2028, and if it misses, the next shot is 2031. "Not a technology problem," Sinacore said. That is either the most reassuring thing a program executive can say about a 60-year-old execution problem, or it is the setup for the next $20 billion in failed programs over time.