Ethereum’s long-distance protocol roadmap could move faster than many expected as AI tools improve, according to Vitalik Buterin. Vitalik Buterin pointed to recent experiments using agentic coding to assemble an ambitious reference client that spans much of Ethereum’s planned 2030-era architecture.
The comment came after developer Jiayao Qi, posting as YQ via X, announced ETH2030, an experimental Ethereum client built to target the network’s draft “2030+” roadmap. The project weighs 702,000 lines of Go, covers 65 roadmap items across 8 phases, has passed 36,126 official Ethereum state tests, and can sync with mainnet through integration with go-ethereum v1.17.0. Qi said the client was built using Claude code in about six days at a cost of about $5,750 and 2.77 billion tokens.
AI could accelerate Ethereum’s roadmap
Buterin called the effort “a very impressive experiment,” but stressed that there are obvious limitations to a prototype built at that speed. “Building something like this in two weeks without even having an EIP requires great care,” he wrote. “It almost certainly has a number of significant bugs, and in some cases it could be a ‘stub’ version of something that the AI never even attempted to create a full version of. But six months ago, even this was well beyond the realm of possibility. What matters is where the trends go.”
For Buterin, that difference was more important than the live demo itself. In his view, AI does more than just reduce development time. It could change the way Ethereum engineers approach warranties. “Perhaps the right way to use AI is to get half the benefits from AI in speed and half in security,” he said. “Generate more test cases, formally verify everything, and do more multiple implementations.”
He tied it directly to the ongoing formal validation work on Ethereum. Referring to the Lean Ethereum initiative, Buterin said one of its collaborators is already using AI to create machine-verifiable proofs of the complex theorems that underpin STARK security. “@leanthereum’s core philosophy is to formally verify everything, and AI is greatly accelerating our ability to do that,” he wrote. “Apart from formal verification, the ability to generate a larger number of test cases is also important.”
ETH2030 itself was presented as a stress test for the roadmap rather than as a potential client. Qi repeatedly described it as a rough draft rather than production software, and argued that its value lies in uncovering difficult engineering problems now rather than years later.
The roadmap implemented in the project aims for a version of Ethereum with over 10,000 TPS in L1, finality in seconds instead of 15 minutes, solo staking of 1 ETH, stateless nodes running on a $7 Raspberry Pi, and over 1 million TPS across L1 and L2. But the experiment also revealed deep coupling between upgrades, from block access lists and gas repricing to PeerDAS, native rollups, and fast finality.
Mr. Qi spoke frankly about the gap. Pure-Go’s crypto implementation is approximately 10x to 100x behind production code, the consensus logic has not been battle-tested on a live beacon chain, and the jump from the current ~5 million gas per second to a goal of 1 billion gas per second remains highly speculative under real-world MEV and contract dependency patterns.
Buterin did not claim that AI would solve these problems. In fact, he cautioned against expecting secure protocols from a single prompt. “There will be a lot of struggle with bugs and inconsistencies between implementations,” he wrote. “But even that wrestling can be done five times faster and ten times more thoroughly.”
More than the headline numbers, that’s what matters to Ethereum researchers and client teams right now. If AI can speed up both implementation and validation, roadmaps may become more than just distant architectural sketches. As Buterin said, people should at least be open to the “possibility” that the Ethereum roadmap could be completed “much sooner than people expected and with a much higher level of security than people expected.”
At the time of writing, ETH was trading at $1,956.
Featured image created with DALL.E, chart on TradingView.com
editing process for is focused on providing thoroughly researched, accurate, and unbiased content. We adhere to strict sourcing standards, and each page is carefully reviewed by our team of top technology experts and experienced editors. This process ensures the integrity, relevance, and value of your content to your readers.
