Because the secondary effects of something becoming more efficient can be harmful, in counterintuitive ways.
E.g. if some process in steel manufacturing is improved to require less power, you might think, yay, there's less power needed so less CO2 released. Good for mitigating global warming. But what might actually happen instead is that it becomes feasible to make more things out of steel because it is cheaper than before, so more energy is used on steel production in total, even though per unit it is more efficient.
The same thing could happen with fusion energy, but in ways we can barely imagine now. What high-power workloads are not even considered now that would be cheap enough with fusion power?
