Two weeks ago, Vitalik followed a man namedYaq made a bet: Can one person use AI programming tools to write the entire Ethereum 2030+ technical roadmap into code?

No one expected this bet to come to fruition so quickly. But now, Yaq is done. It took6days, spent5750US dollars, 36,126an official test, all passed. The code can be compiled, can run, can be synchronized from the test network in 9000 blocks/ seconds, and can be connected to Ethereummainnet.

This project calledETH2030 may be the largest technology roadmap verification in the history of blockchain.

6days70万行

Imagine this scene: a developer, facingClaude Code(Anthropic'sAIProgramming Assistant), wrote 700,000 lines of code in six days.

What is the concept of seven hundred thousand lines? LinuxThe kernel has 30 million lines. ChromeThe browser has 17 million lines. A complete production-grade database has one million rows.

Yaq leaf="">RISC-V CPU, Verkletree. These are the most complex parts of Ethereum's planning.

If you tell this story to a software engineer, their first reaction is usually: How is this possible?

Because what was the past practice? Find a team of ten to twenty people and spend two or three years to write a reference implementation.

ButAIchanged everything. Six days, one person, one prompt word, and a code review process——boom, 700,000 lines of code, able to compile, test, and synchronize the main network.

A super team of one person

This means two key changes.

First, the cost has dropped by an order of magnitude. Projects that once cost millions of dollars now cost just $5750. Secondly, the time to detect problems is earlier.

Before the code for Yaq was released, these potential problems were hidden in the specification document and no one really tried to integrate them. Now anyone can download the code, see the implementation, and ask real questions.

Used by one personAI, it is equivalent to having a super team. YaqVerifies whether a multi-billion ecological technology roadmap is self-consistent. He exposes problems hidden deep within specification documents. He gave the entire community the opportunity to look at the future of Ethereum four years in advance.

But this also exposesAI's fundamental limitation: no matter whatAIcan do, it can never make real strategic decisions. Should Ethereum complete this upgrade before 2030? How to prioritize the roadmap? What risks are acceptable? Only humans can answer these. Only the Ethereum community can answer.

AndYaq's job is to provide a clear framework for this question-answering process. A framework with code, tests, and specific problems.

The bottleneck of the theoretical blueprint

WhenYaqPut all65After all future upgrades were stuffed into the same code system at once, the perfect blueprint on paper began to show challenges at the code level.

Let’s talk about cryptography first. YaqWritten the cryptography part in pureGolanguage (includingBLSsignature,KZGcommitment, etc.).

The result is 10 to 100 times slower than the professional optimized version. This is not because the code was written incorrectly, but because he deliberately chose a quick verification route. If you switch to C or Rust to speed up, the development time will be at least doubled.

Now everyone can clearly see a key question: In a real network, will this speed gap slow down the entire system to a crawl?

Let’s talk about parallel execution. Ethereum wants to allow multiple transactions to run at the same time, and theoretically it can fly. But in reality, transactions often fight with each other and grab resources.

What's even worse is that someone may deliberately create conflicts to destroy the performance of parallel execution. This confrontational risk will only be exposed when the code is actually run.

Finally is the timetable. 65Upgrades are like a chain. If one link is late, the rest will have to wait. This fragility that affects the entire body was previously hidden in documents and no one had really tried it. Now it has become something that everyone can discuss and figure out solutions in advance.

YaqWe did all this not to say that the road map is not working. He just wants to bring these real risks to today in advance, so that the community can see and understand them clearly at the code level and make smarter decisions early.

Four-Year Challenge

2030There are four years left.

EthereumTo complete a generational upgrade within these four years: the processing capacity will increase from500Ten thousandgas/mentioned in seconds10billiongas/seconds, transaction confirmation started from15minutes are reduced to seconds, allowing ordinary people to independently verify the network with 1 ETH, and even allowing Raspberry Pi to run a full node.

At the same time, the migration of post-quantum cryptography must be completed——Nearly 200 million old accounts need to be upgraded,8This stage requires precise coordination, and any delays will have knock-on effects.

Competitors such as Solana, Monad have already delivered high-throughput solutions in the real world. This is no longer just a technical issue, but also a race of time and confidence.

YaqUse6Tianba five years laterEthereum has been completely simulated, bringing the problem from 2030 to today.

The code is open source and tested. Now it is the turn of the engineers, researchers, and cryptographers in the community to take a deep look, find out the real bottlenecks, and decide what needs to be optimized and what needs to be redesigned.

This process may adjust the timeline, but it is much better than2028 only to find out that key designs are simply not compatible.

The future verified in advance

Did Vitalik lose his bet? YaqIt took only six days to complete the seemingly impossible task.

But from a larger perspective, EthereumThe community won. Now we no longer just imagine the future against a paper blueprint, but can download the code, run it, and see what Ethereum will look like in 2030.

We can ask the toughest question in advance: Why is it designed this way? What happens if you experience real stress? Can this hypothesis hold up for four years?

These 700,000 lines of code may become the compass for the next four yearsEthereumand may also be significantly rewritten in community iterations. But at least the direction is clear and the problem is clear.

InAIera, the boundaries of what a person can do are constantly being broken. What we thought was impossible six days ago has now happened. AIallows us to verify the future at an astonishing speed, giving precise questions rather than just abstract discussions.

AndYaqproved that: one person+AI can change the world in five years

We have the tools to quickly simulate the future, which is both an opportunity and a pressure. The opportunity lies in discovering real problems earlier, and the pressure lies in having to make decisions faster——Because with every hesitation, competitors move forward.

The future is being simulated in advance by our own hands.