Efficiency gains. How many "resources" are consumed by a piece of software relative to the "growth" that resulted from it? I can now have a machine generate 100 mediocre pieces of art for $20, two years ago each one would have cost $100 and taken a week for an actual artist to produce.
1. Yes, the resources consumed to make software are smaller, but they're still non-zero and will never be zero. So ultimately you're still limited by finite resources.
2. Many things just can't be implemented in software. I would point to A. Housing and B. Food production as things that are pretty hard to optimise. That isn't to say that we can't optimise them at all. We can build smaller houses or taller buildings. And we can transform energy from fossil fuels into fertiliser that increases yields. But growth in these areas is still very much constrained.
And these things are key human needs.
Like, about 150 years ago an American farm could get about 26 bushels of corn per acre. The last few years, we've gotten 174, about 6x more. And that's in a very favourable agricultural zone, chances are it's even greater in more marginal areas!
Source: https://farmdocdaily.illinois.edu/2022/07/perspectives-on-na...
We see it all the time? Google's revenue is up last quarter despite redundancies meaning they have a lower headcount. Adding robotics to a car production line can massively increase the productivity of a single factory despite needing less headcount, often allowing other factories to shut.
Something like Intellij or Copilot can make the entire tech workforce more efficient _far_ in excess of what it cost to make.
By creating something new that is more efficent. Even if we ignore economies of scale as a source of efficency there is a replacement curve in action which would create growth! A situation where they would immediately scrap the old equipment for the same payload and switch to a new more efficient one is extremely rare.
Ininite growth isn't possible, but I genuinely believe near-infinite growth is.