Ethereum’s engine replacement is about to begin. The proposal recently put forward by Vitalik Buterin is not just an upgrade—it means a fundamental rebuild of the EVM.

Since last year, there has been an unspoken rule within the developer community. Whenever new cryptographic operations are required, instead of modifying the EVM itself, they have used a “workaround” known as precompiled contracts. In other words, they left the foundation untouched and piled on superficial changes. Vitalik has objected to this direction. His argument is simple—Ethereum’s value lies in its general-purpose nature, and if the EVM is inadequate, then a better virtual machine should be built.

Specifically, two major changes are being proposed. First is a reform of the state tree. Today, Ethereum uses a complex “six-branch KeccakMerklePatricia tree,” but the plan is to replace it with a simple binary tree. This would drastically reduce bandwidth during data verification and dramatically improve the efficiency of light clients. There is also consideration of changing the hash function to Blake3 or Poseidon.

The second is an even more ambitious change: replacing the EVM long-term with a RISC-V architecture. In ZK proof systems, RISC-V is already used as the standard language. Vitalik’s logic is straightforward. If the prover is already running on RISC-V, why should the virtual machine speak a different language—requiring a translation layer in between? If you remove the translation layer, efficiency naturally improves.

Taken together, these two changes are said to account for more than 80% of Ethereum’s proof bottleneck. That means that without implementing them, scaling in the ZK era will not move forward.

However, not everyone is on board. Offchain Labs—the core development team of Arbitrum—has published detailed technical rebuttals. Their point is sharp: the distribution format and the proof format do not have to be the same. Just because forklifts are efficient in a warehouse doesn’t mean delivery drivers also need to use forklifts. Offchain Labs proposes a two-layer approach: adopt WebAssembly at the smart contract layer, compile it to RISC-V, and then turn it into proofs. In fact, they have already run prototypes of this approach on Arbitrum.

There is also an even more important concern. In the field of ZK proofs, technical evolution is extremely fast, and RISC-V implementations have only just shifted from 32-bit to 64-bit. If you lock RISC-V into Ethereum L1 at this stage, what happens if a better proof architecture emerges two years from now? Betting on a rapidly changing target is not in Ethereum’s style.

What’s particularly interesting is the timing. Exactly one month ago, Vitalik publicly questioned whether Ethereum needs a “dedicated L2 roadmap.” In response, the L2 camp has instead begun actively pushing for “independence from Ethereum.” The co-founders of OP Labs and the CEO of Polygon have said that the role of L2 is not just scaling—it is about building block space tailored to specific use cases.

In other words, Vitalik’s large-scale changes to the execution layer are merely a technical annotation of a broader trend. Ethereum is regaining control over its core functions, while L2s are finding reasons to exist independently.

The reform of the state tree is already at a high level of maturity, with specific drafts and a dedicated push team behind EIP-7864. Meanwhile, replacing the EVM is still at the “roadmap” stage, and there is a long distance to implementation. However, Vitalik himself has said that Ethereum has already completed a one-time engine swap with The Merge, and that about four more swaps are possible in the future.

The Glamsterdam upgrade is scheduled to be implemented in the first half of 2026, followed by Hegota. The clear main directions are reforming the state tree and optimizing the execution layer.

Ethereum’s story was never about a question of “can it be done or not.” From PoW to PoS, shifting focus from all L1s to rollups, and already demonstrating the ability and courage to disassemble engines at high altitudes—those things have already been shown. What is being moved now is a deeper layer—not adding new crypto features, but digging up the old foundation and re-concreting it.

Is this a deeply planned renewal, or an increasingly complicated bottomless pit? The answer probably won’t be clear until 2027. One thing is certain: Ethereum does not intend to remain “an old system that keeps getting patched” in the ZK era. The debate over how to remove the patches and which engine model to swap in may be more valuable than the conclusion itself.