The historical record has many examples of of states trying to buy out others' nuclear programs. Are such agreements credible in the long run? I develop an infinite horizon bargaining game in which power shifts are endogenous and costly. Three results stand out. First, when the the threat of preventive war is credible, rising states internalize the threat and choose not to build. Second, when the cost to build nuclear weapons is sufficiently large, rising states opt against building without further strategic considerations. Finally, when these incentives fall in a middle range, declining states offer immediate concessions commensurate with the value of the program. Rising states accept immediately. These deals work because dropping the concessions causes rising states to switch to building. This eliminates the surplus created by not spending resources on nuclear weapons, which the states could otherwise split.