(assuming scaling is solved technologically - which is a certainty over the long run, as processors, storage, networks become better).
I don't think so if you had to do everything on-chain. You need to get bitcoin to something like 5000 TPS for it to be used as a currency and valued for it's utility instead of a settlement layer, which it's not very good at because bitcoin is a poor store of value compared to metals. Going the settlement layer route is a recipe for failure in the long run and that's what all on-chain scaling is.
20 years ago (q1 1997), I had a 486/133 (oc @160 roughly around a P90), 16MB RAM, a 4 GB HDD and a 28.8 or 33kbps modem.
In 20 years, processing power/storage/networks have seen gains of 1000x or more.
In 20 years, if we get another 1000x, we'll easily cover 5000TPS and more. I think we'll now see more than 1000x, as AI is introduced in technological research and development - and there is a lot of things that we currently perform in an unoptimized manner.
Going back to 97, if we run Bitcoin back then, it would be considered DOA - a joke that can't even do a tx/sec, and that would demand enormous CPU power, network resources for the home user, filling their HDD fast, etc etc. But now, in 2017, it's "ok" for use.
In 20 years from now it will be able to do way more than visa level txs. On-chain. And that's assuming zero progress in software solutions (which of course won't be the case).
In a way, Bitcoin cannot be judged statically, with the tech of a certain era, as Bitcoin's capabilities can multiply with technological progression. Technological progression is what ensures that Bitcoin will scale. It's not a matter of if, but when. When will people be able to run nodes that handle thousands of tx/sec, from their home connection, with their home pc or other device? Is it 2025, 2030, 2035, 2040? That's the only question...