It's an excellent metric, the amount of applications not viable now due to cost/latency/throughput is vastly bigger than the amount of current use cases. Even current ones do benefit, e.g. it's a great executor subagent.

Energy and intelligence are good too, sure.

Price accounts for computational/architectural efficiency improvements whereas active parameters does not.

What if we used 100% of the brain all the time?

As an end consumer, I don't care about the number of active parameters. I really do care only about the tracked metric (how well does it do the job, and how much does it cost... ideally also with time included, but that wouldn't fit on a 2D chart)

for someone with a limited budget it is actually very important because it makes me less scared to experiment.