The only way I can see your comment being correct is if you're open to redefinitions of “blockchain” which do not require either immutability or the global public ledger. The current blockchains handle a vanishingly-small fraction of global financial traffic using far more equipment, data, and power — there's no magic wand which can allow you to scale up by many orders of magnitude without architectural changes.
That could be solved by splitting into local blockchains so the global network only handles transactions between local banks but that seems like it's basically giving up most of the benefits claimed in blockchain marketing.
The Ethereum roadmap has sharding coming soon, and architectural changes aplenty. Of course in the near term the approach for scaling is rollups, zk-rollups in the mid term and optimistic rollups in the short term (already running).
Those approaches still have immutability and a global public ledger but that ledger no longer lists every single transaction.
Your argument is refuted by the fact the technology to solve the problem you claim will never be solved literally already exists, one version of which has a market cap of almost $10B.[1]
Claiming that a very obvious engineering problem incentivized to being solved is the same as imagining us being incentivized to regress in the way you mention is also weird.
I presumed you were using “blockchain” as a short hand for “distributed fault tolerant system that solves Byzantine generals problem.” If not then I agree blockchains in the literal sense probably are a transitional tech.
That could be solved by splitting into local blockchains so the global network only handles transactions between local banks but that seems like it's basically giving up most of the benefits claimed in blockchain marketing.