[kragen]

I don't know if this is really the reason. Moore's Law started with planar photolithographic silicon semiconductor integrated circuit fabrication (aka making chips) around 1958. We're now in year 50 of planar photolithographic silicon semiconductor integrated circuit fabrication, although now we have environmental laws so we can't just dump hydrofluoric acid into the groundwater, and we have to use X-rays instead of ordinary light, and we worry about smaller dust, stuff like that which makes it a lot harder. But basically we've just been scaling stuff down. We have about another ten years before we reach the physical limits of planar photolithographic silicon semiconductor integrated circuit fabrication, which are about two orders of magnitude away. That means we started about 17 orders of magnitude away.

By contrast, we seem to have reached the fundamental limits of carbon-zinc batteries a couple of centuries ago, and the fundamental limits of ordinary electrolytic batteries in 1990 with the invention of the lithium battery. (Or maybe the zinc-air battery.) The fundamental limits on chemical energy density seem to be about two or three orders of magnitude from where we started, not 17.

In both cases it's possible to do better by using fundamentally different approaches: microturbines, tiny fuel cells, or betavoltaic batteries in the case of batteries; diamond, three-dimensional structures, and molecular assembly in the case of electronics. But all of these have substantial engineering obstacles.