At the physical thermodynamical level, yes. But estimating "calories in" from food intake and estimating "calories out" from exercise apparently have +/-50% error if you take the numbers from the food packaging and from an exercise book respectively, if you actually measure them for a specific person (which is exceedingly hard, almost never done properly, and always takes for granted assumptions that are not proven -- see my answer to DanBC down this thread)
I should have been more precise: "calories in - calories out = calories stored" is a tautology given conservation of energy, as you alluded. But it turns out to be a very inaccurate prediction if weight gain if you assume "calories in"="what's written on the food label", "calories out=what's written in an exercise book about your exercise", and "calories stored=9kcal/g per body weight gained". All 3 are bad measurements, that somehow work out for the majority (60%-90% of the population, depending on how you set your error bars), but break for a minority.
If conservation of energy did not work for 10% of particles, physicists would look for a better description. Somehow, nutrition "experts" are allowed to hand wave the inaccuracies and are given a free pass as scientists. They aren't.
The book itself is lengthy and comprehensive. He cites a lot of research to make his case, and the citations are available in the book.
With that said, I don't think calories in - calories out is wildly inaccurate, and I don't believe Taubes makes that case.