We present a new framework for studying the socially optimal level of generating capacity and public RD&D investments for the electric power sector under decision-dependent technical change uncertainty. We construct a bottom-up stochastic electricity generation capacity expansion model with uncertain endogenous RD&D-based technical change, focusing on solar PV RD&D investment planning for its current prominent role in sustainable energy and climate policy deliberations. We characterize the decision-dependent process of technical change uncertainty as unknown outcomes of RD&D investments that increase the likelihood of success with increasing amounts of RD&D, and calibrate to a novel expert elicitation dataset that accounts for this decision-dependence. The problem is framed as a multi-stage decision under uncertainty, where the decision maker learns and adapts to new information between decision periods. Specifically, our application considers four decision stages, with the decision-maker choosing investment levels for new capacity and solar PV RD&D, while learning about RD&D outcomes that can reduce solar PV costs between each stage. The problem is thus formulated to match the manner in which real-world decisions about RD&D investments in renewable energy are made, and avoids common assumptions of perfect foresight, or uncertainty but no learning, that are often used in practice. Numerical results show that when uncertainty and learning features are both included, the optimal solar PV RD&D investment strategy changes from solutions using other methods. Considering uncertainty and learning results in solar RD&D investment differences as high as 20 percent lower in the first-stage, and 300 percent higher in later stages. We also show that when uncertainty is considered without learning, the fraction of new solar PV capacity investments can be depressed. Overall, this paper shows that it is possible to unify several realistic features of the deployment and development problem for the electricity sector to meet sustainability goals into one framework.