Suppose as you say, all capital grows uniformly (which it doesn't, but that's another issue) at +x% annually. Now suppose that certain demographics experience -(x+n)% of capital transfer, in the form of housing, healthcare, and other expenses. It doesn't matter how fast the "pie" grows as along as capital transfer away from the middle class outstrips the rate of capital growth. You're cherrypicking hypotheticals here.
Unfortunately, most of the layman discourse on the internet seems to blindly follow the mantra of an infinitely expanding economy.
While that may be true on an infinite timescale, in the short and medium runs (scale of economic cycles), the pie is indeed fixed.
This is relevant in the case of disasters as you don't have time to meaningfully increase the number of Generators available etc.
Fluctuations in the mass of the pie at shorter intervals would appear to observers within our universe as discontinuous jumps between discrete quantities of pie. Therefore, during that interval, the pie is fixed.~
Now go away.~
On a geologic timescale, pie has just appeared out of nothing, and is now covering the entire planet. On an infinite timescale, by naive extrapolation, the pie will in the future be expanding faster than the universe itself, to the point where a wafer-thin bite of pie will expand and rupture the esophagus before peristalsis can even push it into the stomach, and persons dying from attempted pie consumption will literally explode in a shower of pie.~
Clearly, the parent post was substituting a hyperbolic term for the longest possible economic timescale, where new technologies may be invented and entirely new supply chains built based upon them. It is easy to claim that at that scale, economic growth will continue without bound for as long as human ingenuity can conceive new ideas.
At shorter scales, the observable size of the pie does sometimes shrink. And if there are periods when it grows, and periods when it shrinks, then logically, there must be periods when it remains the same size, even if those periods are very short.
I would assert that if something is constantly going up, down, or holding steady at intervals that are largely "random", than it could hardly be considered fixed.