Vitalik just shared his thoughts on Ethereum's evolution, and honestly, the changes he's considering for the execution layer are quite ambitious.

As many have noticed, the current state tree isn't really optimal. EIP-7864 proposes a major overhaul by moving from the hexadecimal Merkle Patricia tree to a more efficient binary structure. Specifically, it replaces the current complexity with a binary system using hash functions like Blake3 or Poseidon, which should significantly shorten Merkle branches.

The interesting part is that this binary approach will also group storage locations into pages. This reduces access costs to adjacent storage, and even leaves room for metadata aimed at future state expiration. Basically, it's an optimization designed for the long term.

But that's only half the story. On the virtual machine side, Vitalik envisions something more radical: replacing the EVM with a RISC-V architecture. I know, it sounds huge, but the idea is to improve execution efficiency, simplify ZK proofs on the client side, and make the code easier to implement.

Deployment wouldn't happen all at once. First, replace precompiled contracts, then support the new RISC-V contracts, and finally achieve backward compatibility while gradually replacing the EVM. It's a thoughtful transition, not a sudden change.

These changes really show how Ethereum continues to evolve to address efficiency and scalability issues. It's worth following this direction.