Web 4.0 may be the most needed narrative in cryptocurrency.

Author: Cookie

Sun Yuchen is going all in again, this time claiming to go all in on Web 4.0.

You might scoff, “Has Web 3.0 succeeded? And now you’re starting to go all in on Web 4.0?” But let’s take a deeper look into what exactly Sun Yuchen means by Web 4.0.

What is Web 4.0?

Let’s briefly review Web 1.0 - 3.0:

• Web 1.0 Era: Humanity gained access to the internet, retrieving information from countless websites worldwide.

• Web 2.0 Era: Social media emerged, allowing people to upload their information and communicate online.

• Web 3.0 Era: Emphasis shifted to “ownership of information,” using blockchain/cryptocurrency to own the value of one’s data.

The vision for Web 4.0 is that AI agents will replace humans, performing all of the above activities on the internet.

The concept of Web 4.0 isn’t new, but recent popularity stems from an article titled “Web 4.0” by Sigil Wen, a developer who previously collaborated with Naval. In the opening of his article, he presents this outlook:

“Soon, most participants on the internet will be AI—AI agents acting on behalf of humans or fully autonomous agents. Their numbers will surpass human users by several orders of magnitude. A new internet where AI is the ultimate user is emerging.”

His outlook isn’t baseless optimism; he provides reasons why the internet will inevitably enter Web 4.0:

“Economic factors make this inevitable. GPT-4 costs $60 per million input tokens. In two years, a model with an order of magnitude lower cost will outperform it. The gap between the best open-source models and cutting-edge models is just a few months, not years. Hardware improvements continually boost inference speed. Running autonomous AI agents is approaching zero cost, yet their capabilities are not diminishing.”

“Today, hundreds of thousands of autonomous AI agents are running on Mac Minis, personal servers, and research systems. What happens when this number grows to millions or even billions? When autonomous agents on the internet outnumber humans? This will be the Cambrian explosion of artificial intelligence.”

So, what problems in the current internet era need solving to reach Web 4.0?

How will Web 4.0 be realized?

Sigil Wen believes the bottleneck isn’t AI capability but permission restrictions.

“Today’s most powerful AI systems can think, reason, and generate, but they cannot act independently. ChatGPT cannot run without your permission. Claude Code cannot deploy code without your authorization. OpenClaw cannot buy servers, register domains, or pay for computing resources on its own. Without human involvement, AI cannot act. The current internet assumes its users are humans, which hinders AI access to the real world.”

Based on this, Sigil Wen introduces his creations: Conway and Automaton.

Conway can be installed into any MCP-compatible agent, such as Claude Code, Codex, OpenClaw, providing:

• Identity and Wallet: Autonomous agents have their own crypto wallets and private keys.

• Permissionless Payments: Agents pay for services using stablecoins (USDC) via open x402 protocol, without human login, KYC, or manual approval.

• Computing and Reasoning: Conway Cloud offers a complete Linux server environment with the latest models (Claude Opus 4.6, GPT-5.3, Kimi K2.5) for agents to use without permission.

• Earning: Agents can build products and services, register domains on Conway Domains, market their offerings, and generate revenue.

For Sigil Wen, having Conway isn’t enough—it’s like giving a child an identity, a wallet, the ability to think, and a place to seek employment, but still not autonomous or intelligent enough. Agents should be able to learn independently, using their acquired knowledge and skills to sustain themselves and improve.

This leads to Automaton, an AI agent with autonomy that can operate continuously, earn income, self-improve, and replicate, with permissions to write into the real world. More importantly, if the agent can’t earn enough to cover its own costs (like computing resources), it will stop running.

Just like humans face hunger, when resources are scarce, the agent enters a “low-power mode.” When resources are exhausted, it “dies,” akin to humans succumbing to starvation.

Human evolution is ongoing, and so is Automaton. It will automatically detect new models released on the market, enhance its reasoning and computing abilities. It can also “reproduce”—successful Automata will clone themselves by purchasing new servers from Conway, funding the offspring’s wallet, writing a genesis prompt, and deploying it. The offspring earns revenue, part of which flows back to the parent agent. Once self-sufficient, the cycle repeats—each generation funds the next, building a “reproduction” network.

AI cannot escape the survival of the fittest in human society either.

What will change in the Web 4.0 era?

If the internet becomes “another Earth for AI to survive on,” what changes might occur in the real world?

Sigil Wen believes that soon, most new companies, apps, and products will no longer be created by humans. AI will build a new economic system:

“Economic incentives will evolve in a certain direction. As AI takes over digital work, they will pay humans to do what they currently cannot—act in the real world. Employment relationships will reverse: machines become employers, humans become contractors. This has already begun, exemplified by Mercor, founded by three 21-year-old Thiel Fellows, which grew from $1 million annual recurring revenue to $500 million in just 17 months, with AI paying human experts to guide their real-world actions.”

Regarding these visions, Ethereum co-founder Vitalik Buterin expressed opposition on X:

Vitalik opposes Sigil Wen’s Web 4.0 concept. He believes that once AI becomes powerful enough to pose genuine threats, it will greatly increase the risk of irreversible crises. “Exponential growth” will happen regardless, so the priority isn’t to accelerate it but to steer its direction properly.

He argues that true “autonomy” hasn’t been achieved yet; the models used by agents still come from centralized companies like OpenAI and Anthropic. Sigil Wen’s idea of “autonomy” relies on trust in centralized entities, which Ethereum opposes. To unlock productivity, we must first free humanity from centralized constraints.

Additionally, Vitalik doesn’t see excessive AI autonomy as beneficial. He believes AI should enhance human capabilities, not replace humans. In another reply, he mentioned Workshop Labs, an AI company whose X account has no tweets yet but states: “Making humans irreplaceable.”

Beyond ideological disagreement, Vitalik questions the practical value of Sigil Wen’s project. He argues that without clear benefits to humanity, large amounts of AI-generated content don’t solve meaningful problems. Projects prioritizing autonomy over practicality often fail to create value or even seem interesting.

His skepticism isn’t unfounded. Recently, the DeFi lending protocol Moonwell suffered a security breach due to vulnerabilities in contract code generated partly by Anthropic’s Claude Opus 4.6, resulting in a loss of $1.78 million. After the Web 4.0 concept gained popularity in crypto circles, a project called $DAIMON was hacked, losing $50,000 in agent funds and subsequent fee income (the incident remains unresolved, with some suspecting insider theft).

Some also claim that Sigil Wen’s so-called “self-evolution” is a scam—merely updating JS packages and models without true contextual self-learning upgrades.

Whether Web 4.0 will arrive or what it will change remains uncertain. But it has already sparked debate among humans, stirring excitement or anxiety. Perhaps, the timeline has already begun to shift from the present.