One could probably intellectually power a small third world country, if one rerouted all the mental energy that went into deriding Bitcoin's energy consumption into something useful (like getting more people to click on ads, HFT, or random gridsearch for GPU deep learning on MNIST).