NASA paid for milestones, not costs, and got rockets at a tenth the price

Since the Shuttle era, NASA had developed vehicles mostly through traditional cost-plus contracts: the government pays a contractor's costs plus a guaranteed margin, whether or not the vehicle flies on time or on budget. 7 to $4 billion just to develop a rocket in the Falcon 9 class. After the 2003 Columbia disaster and the decision to retire the Shuttle, NASA needed to get cargo, and eventually crew, to the space station without either flying aging hardware indefinitely or committing to another decade-long, multibillion-dollar traditional program for every new vehicle.

The obvious fix — a normal Federal Acquisition Regulation contract with tighter cost controls and better contractor selection — leaves the underlying problem untouched: a cost-plus contractor is still paid the same whether it comes in on budget or triples it, and the FAR's own compliance and reporting requirements add real cost independent of how hard the actual engineering problem is.

NASA used 1958 Space Act authority to pay fixed sums only on verifiable milestones, making companies absorb overruns and keep underruns as profit.

Paying only on milestones removes the contractor's incentive to inflate costs, since overruns are their loss.

Fixed sums let contractors keep underruns as profit, motivating efficient engineering.

Objective milestones prevent payment for paper progress, ensuring real capability exists.

NASA's legal authority enabled bypassing FAR's cost-plus rules, allowing a fundamentally different incentive structure.

When you can't supervise closely, pay only for verified results, letting the vendor keep underruns and absorb overruns.

NASA's own cost-assessment data (Edgar Zapata, NASA Kennedy Space Center, 2017) puts SpaceX's actual Falcon 9 development cost at roughly $300 million, against NASA's own cost-plus-based model predicting $1.7 to $4 billion for an equivalent vehicle developed the traditional way. Across the full COTS cargo program, NASA's total investment of roughly $953 million delivered two new launch vehicles and two cargo spacecraft, versus an estimated $3,977 million for the traditional-contract equivalent — about a tenth of the cost.