I think L2 has many more uses than just reducing fees. For example, decentralized exchanges, NFTs, games, and many things that are not considered transactions should be there. Here, the fees will be constantly low with the possibility of being able to be scalable in unnecessary directions.
Let's not forget about L3s. It's the next big thing. Layers built on top of other layers. Just like how the Internet works today. Even Sony built its own L2 network on top of the Ethereum blockchain. Highly-scalable L1s like Solana and BSC might need L2s if they get plagued by network congestion. There's no such thing as "infinite scalability", though. If developers want to keep the main chain decentralized, moving certain transactions off-chain (L2) would be the way to go.
I think Polygon will still be relevant in the future. It's already fast, cheap, and highly-compatible with the EVM. It's even better than separate L2 solutions such as Arbitrum and Optimism. I've read that developers will be integrating ZKPs on the Polygon chain. It will make Polygon more scalable and private. The change from MATIC to POL is only a rebranding. But if the project keeps innovating, it might reach the top someday. Who knows if POL goes as far as surpassing SOL in market cap?
