He suggests that most of the energy Bitcoin consumes generally comes from places that consistently produce an excess of energy that otherwise goes to waste. Miners, seeking to maximize profit, gravitate to areas where energy is cheap because supply > demand. I'm not sure what percentage of the energy could be considered "excess."
His other point was that the discussion of Bitcoin's energy consumption should be framed in the context of how much value it brings to the global community. Expecting a valuable global decentralized asset to be entirely without energy costs is unrealistic, it's just a matter of how much energy is feasible. Depending on how valuable Bitcoin is deemed to be, consuming the same amount of electricity as a small country might be justified in the short term, but should clearly be addressed in mid term.