NSF’s strategy of funding early-stage research with unknowable outcomes was essential to making the LLVM academic research project and infrastructure possible.