We propose one of the first dynamic models of the optimal transition from fossil fuels to renewables in electricity generation that takes into account the variability and intermittency of renewable energy sources as well as storage. We take as an example solar energy, which is variable (no sun at night) and intermittent (few or no sun at day when there are clouds). We show that when the clouds phenomenon is not too severe, intermittency can be safely ignored and the planner just needs to take into account the deterministic variability of the renewable source. In this case, the optimal transition consists in using fossil fuels at day and night and complement them by solar electricity at day while investing to build up solar capacity; then abandoning fossils at day and keeping them for night electricity generation only; then, when solar capacity is large enough, starting to store electricity; and finally abandoning totally fossils when the carbon budget is exhausted. However, if the cloud problem is severe, intermittency matters a lot and precaution requires to start storage earlier, before fossils have been abandoned at day. We show that renewable electricity generation and storage are complement: absent storage devices, the long run solar capacity is smaller, and so is electricity consumption. We finally provide a quantitative illustration for the case of the Spanish energy transition. We show that in Spain intermittency can be safely ignored. We compute the carbon value corresponding to a 2 C carbon budget, the dates at which storage starts, and the path of investment in solar capacity.