The N^2 problem of Bitcoin and Ethereum is solvable through sharding. I agree that on-chain scaling is impossible (and the two projects differ in their taste for decentralization vs scalability at their core), but the build out of workable sharding solutions with reasonable trade-offs and a workable topology are underway by both camps. It's just going to take longer than I think a lot of people who bought in with the pay-per-service use case would prefer.

In the case of chat, I'm convinced that the ability to register data in a shared, immutable store like a blockchain with many stakeholders of disparate interests will be valuable to that end even before microtransactions start being sent with every packet on the internet.