On September 17, 1955, a modified B-36 bomber lifted off from Carswell Air Force Base in Fort Worth, Texas, with a live 1-megawatt air-cooled nuclear reactor humming in its bomb bay. Over 47 test flights through 1957, the NB-36H "Crusader" proved that a reactor could operate safely aboard an aircraft, shielded by 12 tons of lead, rubber, and water surrounding a lead-glass cockpit four feet thick. The reactor never powered the engines — it was a flying testbed for the real prize: the WS-125, a nuclear-powered strategic bomber with effectively unlimited range.

The program, called Aircraft Nuclear Propulsion (ANP), ran from 1946 to 1961 and consumed roughly $1 billion (about $11 billion today). General Electric and Pratt & Whitney pursued two competing engine designs: GE's direct-cycle turbojet, where reactor air passed straight through the turbines (efficient, but radioactive exhaust), and P&W's indirect-cycle using molten-salt heat exchangers (cleaner, but heavier). GE's HTRE-3 test reactor at the Idaho National Reactor Testing Station successfully ran a modified J47 jet engine on nuclear heat in 1956.

Kennedy killed it on March 28, 1961, in a single budget message. The reasons were brutal and correct for the era:

• Shielding weight: protecting a crew required so much mass that range advantages evaporated. Designs ballooned to 600,000+ lb gross weight.
• Crash risk: a downed nuclear bomber would scatter fission products across whatever it hit. The 1966 Palomares B-52 incident later confirmed the political nightmare.
• ICBMs made it pointless: by 1961, Atlas and Minuteman could deliver warheads in 30 minutes. Why need a bomber that loiters for weeks?
• Materials limits: 1950s superalloys couldn't sustain the 1,800°F+ turbine inlet temperatures needed for efficient direct-cycle operation.

Why 2026 changes the equation — but for a different mission. Nobody wants nuclear bombers. But the technology stack underneath ANP is suddenly relevant for uncrewed long-endurance platforms and deep-space propulsion:

• No crew, no shielding: an autonomous platform needs only a "shadow shield" protecting the avionics — cutting shielding mass by 80%+. DARPA's DRACO program (nuclear thermal rocket, scheduled for 2027 demo with Lockheed and BWXT) is essentially picking up where ANP left off.
• HALEU and TRISO fuel: high-assay low-enriched uranium in tri-structural isotropic particles can survive 1,800°C without releasing fission products — solving the crash-dispersal problem that doomed the direct cycle.
• Molten-salt heat exchangers: the indirect-cycle approach P&W struggled with is now mature thanks to Gen-IV reactor research. Kairos Power and TerraPower have demonstrated FLiBe loops at relevant temperatures.
• Additive manufacturing: GE's 1950s heat exchangers were brazed plate stacks that leaked. Modern SLM-printed Inconel 718 monolithic exchangers handle higher pressures with zero joints.
• Refractory composites: SiC-SiC ceramic matrix composites tolerate 2,400°F turbine inlets — making efficient closed-cycle Brayton turbines finally practical.

The NB-36H itself was scrapped in 1958 at Fort Worth. The HTRE-3 reactor sits outside the EBR-1 museum in Idaho, quietly radioactive. But the engineering data — thousands of hours of operating experience with airborne reactors — is declassified and sitting in DOE archives. Any serious nuclear-electric or nuclear-thermal aerospace program today is, whether they admit it or not, building on ANP's foundation.