This is an empty argument; obviously nothing can be emission-free for as long as we run our cars, planes and ships on (non-captured-CO2) carbon fuels; but if the source of our energy is net emission-free (i.e. if you used the energy to build and decomission plants and mine and process nuclear fuel, you'd still have energy left over), then we can slowly move the rest of the society over to electricity as well. Without "almost-but-not-really-emission-free" power sources, that's simply something we cannot do.

Nuclear power is actually free of CO2 emissions. It's easy to see: one could use the electricity to make synthetic fuels from air and water. (Capture of CO2 sounds like a difficult problem, but ammonia and hydrazine would definitely work.) If the supply chain uses synfuels, nuclear power is emission free.

However, as long as a coal fired power plant is operating somewhere, you achieve a better result by using fossil fuels for the supply chain, feeding the electricity into the grid and saving more CO2 emissions elsewhere.

By the way, Sovacool's basic argument is that nuclear isn't CO2 free because the grid isn't nuclear enough. Something along the lines of "if we built more nuclear plants, we'd have to build more old enrichment plants and more coal plants to power them". In other words, he's too stupid or too crooked to correctly compute total derivatives.