The consolidation wasn't market forces working themselves out. It was announced, at a dinner, on purpose. In July 1993, Defense Secretary Les Aspin and his deputy William Perry invited the chief executives of major defense contractors to a dinner at the Pentagon, and told them plainly: the defense budget, already sliding since the Cold War's end, was going to fall much further, and the 51 prime contractors then splitting that shrinking pie could not all survive it. Martin Marietta's CEO, Norman Augustine, coined the dinner's nickname on the spot: the Last Supper, because not everyone at the table was going to make it.^[1]

They didn't. Northrop bought Grumman in 1994. Lockheed merged with Martin Marietta in 1995, a deal that drew no objection from either the Justice Department or the FTC and became the template every merger after it pointed to. Boeing bought McDonnell Douglas in 1997, a $13.3 billion deal that combined the only two remaining US manufacturers of commercial airliners into one -- which is the specific reason there is exactly one American company that builds passenger jets at Boeing's scale today, not a competitive field of them. By the end of the 1990s, the number of American aerospace and defense prime contractors had gone from 51 to 5.^[1] Boeing and Lockheed Martin are not "two big airplane companies" in the way Coke and Pepsi are two big soda companies. They're two of five survivors of a government-orchestrated culling, in an industry the government decided in advance could not be allowed to shrink past a certain point.

What that survival actually buys the country shows up most clearly in a program almost nobody was allowed to know about while it was happening. In 1959 the CIA gave Lockheed's Skunk Works division, run by the engineer Clarence "Kelly" Johnson, a contract for an aircraft that didn't exist yet in any material sense: something that could fly over the Soviet Union at more than three times the speed of sound, above 85,000 feet, invisible enough to survive it.^[2] Ordinary aluminum melts at Mach 3 airframe temperatures, so Lockheed built the A-12 Oxcart -- first flight April 26, 1962, out of the secret Groom Lake facility now known as Area 51 -- out of more than 90% titanium, a metal the US didn't have nearly enough of at the purity the airframe needed.^[2] The CIA's fix was to buy the titanium through shell companies, under a cover story that it was destined for commercial pizza ovens, from the country with the largest known titanium reserves in the world at the time: the Soviet Union.^[2] A dozen Oxcarts were built for operational service. The plane meant to spy on the USSR flew on metal the USSR sold the US without knowing why. That's not a factory's output. That's an accumulated, non-transferable body of engineering knowledge that existed nowhere else on Earth in 1962, and arguably still doesn't move easily between companies today.

The flip side of that -- what happens once a capability like that actually lapses -- has its own recent, well-documented example. The Air Force originally planned to buy 750 F-22 Raptors, Lockheed's next stealth fighter generation after the Oxcart's descendants.^[3] Cost pressure and a 2009 Pentagon decision that a fifth-generation fighter wasn't worth the money at that scale capped the order at 187.^[3] The last F-22 was completed at Lockheed's Marietta, Georgia line in December 2011. The tooling and remaining parts went into storage, officially in case the line ever needed to restart. By every serious account since, it can't be restarted except at prohibitive cost -- not because Lockheed forgot how, but because the specialized workforce, the supplier relationships, and the tooling calibration that made 187 airplanes possible don't survive years of sitting in a warehouse; a 2016-2017 Air Force cost study put a restart at roughly $50 billion for 194 more aircraft.^[3] Once, the same country covertly imported enemy titanium to build a dozen airplanes nobody thought were possible. A generation later, it couldn't restart a production line it had run for years, because it let the underlying capability go quiet.

Which is the real reason Boeing's 2019 crisis didn't touch its defense business at all. Two 737 MAX crashes -- Lion Air Flight 610 in October 2018 and Ethiopian Airlines Flight 302 in March 2019 -- killed 346 people; the FAA grounded the entire fleet on March 13, 2019.^[4] In the middle of that same year, the Pentagon's budget kept flowing to Boeing's tanker and fighter lines -- the KC-46 refueling tanker, and F-15 and F/A-18 orders -- both built only by Boeing. That's not the Pentagon looking past a bad year out of loyalty. It's the practical fact that there is no second American company positioned to build tankers and fighters at that scale on short notice, any more than there's a second company positioned to build a passenger jet at Boeing's scale. The 1993 dinner didn't just shrink the industry to five companies. It removed the redundancy that would have let the government simply route around a failing one.

Why does this matter? Every material in this story -- the Oxcart's titanium, the F-22's airframe alloys, a 737's fuselage aluminum -- runs through the same domestic metals question steel and aluminum tariffs are fighting over right now, and the underlying logic is the same one that kept a civilian auto industry standing since 1940 and four separate shipyards alive for four separate reasons. Aerospace just ran the experiment furthest: the government didn't wait for the market to decide how many companies could build a fighter jet, it convened the dinner and decided in advance, then watched the F-22 line prove what happens to whatever's left outside that circle. Five companies isn't a market outcome. It's a floor the government set on purpose, in 1993, and has been managing around ever since.