How SpaceX Beat NASA's Own Contractors

A two-year-old company with no rocket to its name forced NASA to rewrite how it funds spaceflight, then beat the contractors who were supposed to win easily.

In February 2004, NASA gave a quarter of a billion dollars to a rocket company that had already gone bankrupt. Kistler Aerospace had been developing a reusable launcher called the K-1 since 1994. Funded partly by NASA itself, which makes the bankruptcy filing the year before the award landed a slightly awkward detail for everyone involved. There was no open competition. The contract simply appeared, worth $227 million, and the aerospace establishment did not seem especially surprised by that.

SpaceX was two years old at the time, had not yet launched anything, and was not invited to bid. Musk protested anyway, arguing in an April 2004 investor update that handing a quarter of a billion dollars to a company that had not completed a working prototype in eleven years, without so much as a competitive process, sent a fairly clear signal about how NASA still did business. He was, on paper, a nobody objecting to how the grown-ups allocated money. The kind of protest that usually goes nowhere.

This one did go somewhere. The Government Accountability Office agreed to hear SpaceX's protest, and before a ruling was issued, NASA withdrew the Kistler contract in the summer of 2004. A two-year-old company with no flight history had just forced NASA to unwind a quarter-billion-dollar award to an established contractor. I don't think anyone at Kistler saw that coming, and I'm not sure NASA did either.

How the SpaceX NASA Contract Almost Went to Kistler Instead

What NASA did next mattered more than the reversal itself. Back at the drawing board, and still needing a way to resupply the International Space Station once the Shuttle retired, the agency designed something genuinely unlike its usual procurement process. It was called Commercial Orbital Transportation Services, COTS for short, and the mechanism underneath it was the interesting part. Instead of a traditional cost-plus contract, where a contractor gets paid for whatever it spends plus a guaranteed fee regardless of whether the thing ever flies, COTS paid out only when a company hit a specific technical milestone. Design review complete, payment released. Engine test passed, payment released. Nothing delivered, nothing paid.

Cost-plus is the model that built the Shuttle programme, and the Shuttle programme is a useful comparison here because of how badly the incentives point in the wrong direction. Adjusted for inflation, it cost NASA in the region of $196 billion over its lifetime. A contractor operating under cost-plus faces weaker incentives to finish early or under budget, since finishing means the payments stop. COTS inverted that completely. Miss a milestone and the money simply does not arrive.

NASA opened the new competition in 2005 and selected two winners in August 2006. SpaceX got up to $278 million. Rocketplane Kistler, the same company from the withdrawn 2004 deal, got up to $207 million under the new milestone structure, which is either NASA being scrupulously fair or NASA hedging its bets. Possibly both. Either way, the company at the centre of the original protest had just been handed a second chance, this time under rules built to actually test whether it could deliver.

It could not. Rocketplane Kistler failed to raise the private capital NASA required by its July 2007 deadline and had completed only a handful of milestones by then, worth $32.1 million against the $207 million on offer. NASA terminated the agreement two months later. This is, I think, the detail worth pondering. Under the old system, Kistler would likely have kept receiving cost-plus payments regardless of progress, the same pattern that let Shuttle costs run for decades without a working alternative in sight. Under COTS, the underperformance simply had nowhere to hide. The money that would have gone to Kistler was redirected, with Orbital Sciences Corporation picking up $170 million of it in 2008 to develop the Antares rocket and Cygnus spacecraft.

Falcon 1's Fourth Try

SpaceX, meanwhile, kept losing rockets it desperately needed to work. Falcon 1 had gone through three consecutive failures by August 2008, and the COTS milestone payments were, by the company's own later admission, part of what kept it solvent while Falcon 1 kept not reaching orbit. The fourth Falcon 1 launch succeeded on 28 September 2008. Three months later, NASA awarded SpaceX a $1.6 billion contract to fly cargo to the space station, with Orbital Sciences receiving $1.9 billion for the same purpose. A company that had been three failed launches from insolvency was, within a single quarter, holding one of NASA's largest commercial contracts.

By the time COTS wound down in 2013, NASA's own accounting made the comparison hard to avoid. Roughly $800 million in COTS investment had produced two new launch vehicles and two new cargo spacecraft. For scale, that is less than half of one percent of what the Shuttle programme cost across its lifetime, and considerably less than Orion's development costs alone under a traditional cost-plus arrangement. Two different funding models, tested more or less side by side, and it is not really a contest.

That's not to say cost-plus is necessarily a bad model. Some programmes, deep-space missions with no commercial market behind them, genuinely need a contractor willing to absorb technical risk NASA cannot price in advance. A fixed-price company has no incentive to take that on, and you can hardly blame it. What COTS proved was narrower than "commercial always beats government," and more useful for it. For a well-defined, repeatable task like getting cargo to a station that already exists, competition and milestone payments beat guaranteed fees. The traditional contractors did not lose because they were incompetent. The incentive structure they'd spent decades building around simply stopped matching the job in front of them, and nobody had bothered to check.

SpaceX has since flown crew to the station and landed boosters upright on drone ships, which still looks like science fiction no matter how many times you watch it on YouTube. It's now building a rocket the size of Starship. None of that happens if that 2004 protest goes nowhere and Kistler simply keeps its quarter of a billion dollars. Worth remembering the next time someone frames SpaceX's rise as pure engineering brilliance. Some of it was a two-year-old company reading a contract nobody else thought was worth challenging.

Frequently Asked Questions