[Long English Thread] Ethereum's New Journey: Towards 10,000 TPS and the Ultimate Vision of "ZK Everything"

Chainfeeds Guide:

Ethereum is preparing for the most significant architectural shift since its inception: replacing the EVM with RISC-V. The reason is simple—in a ZK-first future, the EVM has become the biggest bottleneck.

Source:

https://x.com/0xJaehaerys/status/1960051628129865918

Author:

jaehaerys

Opinion:

jaehaerys: Ethereum stands at the most important architectural turning point since its genesis: replacing the Ethereum Virtual Machine (EVM) with RISC-V. The fundamental driver behind this change is that in the era of zero-knowledge proofs (ZK), the EVM has become the largest bottleneck. Currently, zkEVM operates by first recompiling the EVM interpreter into RISC-V and then generating proofs, a process that results in a 50–800x performance loss. As other components such as hash functions have already been optimized, block execution proofs have gradually become the most time-consuming part, and the inefficiency of the EVM is fully magnified. Vitalik pointed out that instead of wrapping another EVM layer on top, it’s better to expose RISC-V directly and eliminate redundant interpretation layers. Meanwhile, the “precompiled contracts” that Ethereum has long relied on have also proven to be a source of technical debt. These hardcoded functions increase protocol complexity and consensus risk, and every addition requires a hard fork, slowing down innovation. Vitalik has clearly stated that new precompiles should be paused in favor of RISC-V implementations. Moreover, the EVM’s 256-bit stack architecture is extremely inefficient, especially when integer calculations common in ZK scenarios only require 32/64 bits, amplifying overhead by 2–4 times. In summary, the three major issues of performance bottlenecks, complexity debt, and outdated architecture make “rebuilding Ethereum’s execution layer with RISC-V” an urgent choice. RISC-V is not a specific product, but an open, modular instruction set standard. Its minimalist core contains only about 47 basic instructions, offering security and verifiability far beyond the bloated EVM; its open ecosystem also means developers don’t need to build compiler toolchains from scratch—mainstream languages like Rust, C++, and Go can be compiled directly to RISC-V, greatly lowering the entry barrier for Ethereum developers. More importantly, the ZK ecosystem has almost reached a consensus: 9 out of 10 zkVM projects have chosen RISC-V as their underlying ISA, proving that this standard has become the de facto industry solution. Compared to the obscure and ambiguous “Yellow Paper,” RISC-V has a formal, machine-readable SAIL specification, providing a foundation for mathematically rigorous correctness verification. At the same time, RISC-V’s “privileged mode” offers hardware-level isolation for trusted execution environments—user-mode contracts must call kernel mode via the ECALL instruction, building a security boundary stronger than the EVM’s software sandbox. The migration path will be divided into three phases: the first phase replaces precompiles with RISC-V in low-risk scenarios; the second phase enters a dual-VM era, where EVM and RISC-V contracts can call each other; the third phase is the “Rosetta strategy,” making the EVM a contract running on RISC-V for long-term compatibility. This gradual evolution ensures stability while paving the way for ultimate protocol simplification. The EVM→RISC-V transition is not just an execution layer upgrade, but will reshape the entire Ethereum ecosystem. For Rollups, optimistic solutions (Arbitrum, Optimism) rely on L1 EVM for fraud-proof re-execution; once the EVM is replaced, their models will face complete restructuring or decoupling risks. ZK Rollups, on the other hand, will enjoy strategic dividends: most projects already use RISC-V internally, and L1 and L2 execution environments will be natively unified, enabling “native Rollups” with efficient, seamless settlement on L1 and lower proof costs. For developers, languages like Rust, Go, and Python will run directly on L1, greatly expanding the boundaries of smart contract development; for users, proof costs may drop by a hundredfold, and Ethereum will achieve “Gigagas L1” with tens of thousands of TPS. In the longer term, this transformation aligns with the “Lean Ethereum” vision: streamlining consensus, data, and execution layers, allowing Ethereum to evolve into a minimalist, verifiable trust layer for the internet. Vitalik’s proposal that “the endgame is ZK-snark everything” is no longer a fantasy, but a blueprint moving toward reality with RISC-V. This means that the future of Ethereum will no longer be a traditional “smart contract virtual machine,” but a globally verifiable, concise, and transparent trust cornerstone. [Original text in English]