I recently noticed that Vitalik Buterin has started speaking frankly about a problem that Ethereum developers have long ignored: over-reliance on shortcuts instead of fixing the fundamentals. In fact, developers have avoided improving the EVM virtual machine, and instead add "pre-compiled contracts" whenever they need a new cryptographic operation. But Vitalik has finally decided to confront this issue head-on.

The plan involves two radical modifications. The first targets the Ethereum state tree — think of it as the network’s indexing system. The problem is that it has become "bloated" due to its current structure. Vitalik proposes replacing it with a simpler binary tree under EIP-7864. The result? The length of a Merkle branch will be reduced to a quarter of its original size, meaning a significant decrease in bandwidth requirements for light clients.

But the most daring part is the second: replacing the EVM with RISC-V in the long term. RISC-V is an open-source instruction set already used in most ZK proof systems. The logic is simple: why use a different virtual machine language and add a translation layer between them? Removing this layer will automatically improve efficiency.

The plan moves in three phases: first, running pre-compiled contracts on a new machine and rewriting 80% of them; second, allowing developers to deploy new contracts directly with parallel execution alongside the old EVM; third, retiring the EVM but rewriting it as a smart contract — old contracts don’t need to change, only the engine does.

Vitalik provided an important figure: the state tree and the virtual machine together account for more than 80% of the proof obstacle. Without these changes, scaling in the ZK era will be just spinning wheels.

But not everyone agrees. The Offchain Labs team (Arbitrum) published a detailed technical response opposing the idea of making RISC-V a unified standard. Their view: RISC-V is good for proofs, but not suitable as a "contract delivery format." They suggest WebAssembly (WASM) instead — it runs efficiently on regular hardware and has mature safety verification mechanisms. Most importantly, they built a prototype on Arbitrum demonstrating that WASM as a delivery layer with RISC-V for proofs can operate independently without interference.

What’s interesting is that this reflects a broader shift in the ecosystem: L2s are starting to seek their own identity beyond just helping Ethereum. Some see the real challenge not as scalability, but as creating unique block spaces for real-world scenarios.

Vitalik admitted that replacing the EVM still lacks widespread compatibility. The state tree fix is more mature, but RISC-V is still on the "roadmap." However, he issued a surprising stance: Ethereum has already replaced its engine once during The Merge, and it can do so about four more times.

The implementation of the Glamsterdam upgrade is expected in the first half of 2026, followed directly by Hegota. Fixing the state tree and improving the execution layer are the two confirmed axes.

The real question: is this a well-thought-out revision or an endless trap that gets more complicated with each fix? We may only find out by 2027. But one thing is certain: Ethereum does not intend to be an outdated system that is only fixed in the age of ZK and modern crypto. The discussion itself may be more valuable than the final outcome.