Bitcoin's Proof of Work algorithm has linear scaling. Basically 100,000 nodes all trying to build consensus using Nakamoto PoW can do it in O(100,000) messages. The scaling factor on the number of miners is not what is a bottleneck here.

The main bottleneck for cryptocurrencies is that every single node has to validate every single transaction. So your global throughput is effectively limited to what a single node can process.

Technologies like STARKs can improve these bottlenecks without introducing new trust layers or trust assumptions but still carry data availability requirements which once again require every node to have all the data, even if they don't have to actually process all the data.

There are additional techniques you can use to minimize the data availability impact but then you start introducing trust assumptions again.

It's a tricky problem and there's a good reason none of the major chains have adopted a solution, but it's not as dire as your post suggests, nor does your post highlight any of the fundamental issues at play.