In the long term, Ethereum is a great thing. However, at this point Ethereum is trying to solve multiple problems at once while not being great at solving any of them.
For instance, we have several major blockchain applications: transactions processing (sending money), DeFi (basically any financially-related apps), NFTs and web3 (or the decentralized web).
I think, at this stage Ethereum is not good at any of these. The transaction processing is slow, the fees for transactions are high — these issues will prevent Ethereum from reaching mass adoption.
I think what Solana does great is high performance low fee processing that would allow all the above applications to scale massively and reach mass adoption.
The most important distinction is that Solana focuses on these two things: transactions price and performance and it does a pretty great job today, not in future. Additionally, the capacity of the network should be more than enough to handle the load for the foreseeable future: https://www.youtube.com/watch?v=ZyIxWutfZ-U.
So even though Ethereum has network effects, and Solana doesn't, with the current trajectory it won't take much time for ecosystem to migrate to Solana.
I think even after PoS Ethereum migration we still have a problem of sharding which has to be implemented, migration of the ecosystem to shards, and basically all the things in between which is a huge task. However, even if Ethereum succeeds at that, it will still be at a point where Solana is already today. Web3 applications are so far away that it might take another few years to implement that.
I think Solana will beat Ethereum in the short run in transactions, DeFi and NFT. It doesn't target web3 but that's totally fine, no one is really solving that currently anyway.
When it comes to decentralization, it's tricky. I mean yes, you could run an Ethereum node on a laptop, however this node won't be able to mine transactions, you'll have to join a pool. In Solana you have to buy a $5k machine, yes, but no one's stopping you from doing that. Additionally, in Ethereum you have to stake 32ETH to be able to mine transactions, so I don't think that Ethereum is so much ahead of Solana in that aspect. I think this might change when sharding will be actively adopted. However, it might take 2 years which is a lot of time.
Arbitrum has already shipped an Ethereum shard however it is still PoW based and from what I understood is still a bit clumsy. It will take time for ecosystem to adapt. Shards don't matter if apps don't use them. There's also ecosystem fragmentation, where there are competing shards providers and now there are apps that are not composable, etc. For the time it would take Ethereum to adopt sharding and migrate apps to them, Solana could basically take over Ethereum in the areas I described above. At this point Ethereum could still tackle web3, but this feels so far away.
For me personally, Ethereum is much closer to my beliefs philosophically, but as it stands today its vision is not there yet in terms of actual implementation: it is still centralized in a similar way to Solana, it is not environmentally friendly and it is not fast or affordable preventing it from gaining mass adoption. However, long term Ethereum has a good chance to have a truly decentralized network that is practical and affordable to use and maintain for everyone, powering the next generation of web that belongs to the people.