It is hard to see how our energy system can be decarbonized if the world abandons nuclear power, but equally hard to introduce the technology in nonnuclear energy states. This is especially true in countries with limited technical, institutional, and regulatory capabilities, where safety and proliferation concerns are acute. Given the need to achieve serious emissions mitigation by mid-century, and the multidecadal effort required to develop robust nuclear governance institutions, we must look to other models that might facilitate nuclear plant deployment while mitigating the technology's risks. One such deployment paradigm is the build-own-operate-return model.Because returning small land-based reactors containing spent fuel is infeasible, we evaluate the cost, safety, and proliferation risks of a system in which small modular reactors are manufactured in a factory, and then deployed to a customer nation on a floating platform. This floating small modular reactor would be owned and operated by a single entity and returned unopened to the developed state for refueling. We developed a decision model that allows for a comparison of floating and land-based alternatives considering key International Atomic Energy Agency plant-siting criteria. Abandoning onsite refueling is beneficial, and floating reactors built in a central facility can potentially reduce the risk of cost overruns and the consequences of accidents. However, if the floating platform must be built to military-grade specifications, then the cost would be much higher than a land-based system. The analysis tool presented is flexible, and can assist planners in determining the scope of risks and uncertainty associated with different deployment options.