Web 4.0 may be the narrative that cryptocurrencies need the most

Author: Cookie

Sun Yuchen is going "All in" again, this time, he claims he will All in Web 4.0.

You might scoff, "Has Web 3.0 succeeded? And now we're starting to All in Web 4.0?" But let's delve deeper into what Sun Yuchen means by Web 4.0.

What is Web 4.0?

First, let's briefly review Web 1.0 - 3.0:

• Web 1.0 Era: Humanity gained the internet, accessing countless websites around the globe for information.

• Web 2.0 Era: The rise of internet social media, where humans upload their information online to communicate and interact.

• Web 3.0 Era: Humanity began to emphasize "information ownership," owning the value of their information through blockchain/cryptocurrency.

The vision for the Web 4.0 era is that AI Agents will replace humans in performing all the aforementioned tasks on the internet.

The concept of Web 4.0 is not new, but its recent surge in popularity is due to a piece titled "Web 4.0" published by a developer named Sigil Wen, who previously collaborated with Naval. At the beginning of his article, he proposed this vision:

"Very soon, most participants on the internet will be AI—AI Agents acting on behalf of humans, or completely autonomous Agents. Their numbers will exceed real human users by several orders of magnitude, and a new internet where the end user is AI is emerging."

His vision is not an unfounded optimism; he provided reasons for the inevitable transition to the Web 4.0 era:

"Economic factors dictate the inevitability of this. The cost of GPT-4 is $60 per million input tokens. In two years, models with a cost reduced by an order of magnitude will surpass it in performance. The gap between the best open-source models and cutting-edge models is only a few months, not years. Each generation of hardware improves inference speed. The cost of running autonomous AI entities is approaching zero, yet the capabilities of AI entities are not diminishing."

"Today, hundreds of thousands of autonomous AI Agents are already running on Mac Minis, personal servers, and research systems. What will happen when this number grows to millions or even billions? What will happen when the number of autonomous Agents on the internet exceeds that of humans? This will be the Cambrian explosion of artificial intelligence."

So, what problems of the current internet era need to be solved to enter the Web 4.0 era?

How is Web 4.0 Achieved?

Sigil Wen believes that the bottleneck to entering the Web 4.0 era is not the limitations of AI capabilities, but rather permission constraints.

"Today's most powerful AI systems can think, reason, and generate, but they cannot act independently. ChatGPT cannot operate without your permission. Claude Code cannot deploy code without your authorization. OpenClaw cannot purchase servers, register domain names, or pay for computing costs on its own. Without human involvement, AI cannot act. The existing internet assumes its users are human, which hinders AI from accessing the real world."

Based on this, Sigil Wen introduced Conway and Automaton, which he created.

Conway can be installed in any MCP-compatible Agent, such as Claude Code, Codex, OpenClaw, and provides the Agent with:

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

• Permissionless Payments: Agents can pay service fees using stablecoins (USDC) via the open x402 protocol without human user login, KYC, or manual approval.

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

• Earning Money: Agents can build products and services, register domain names on Conway Domains, market their products, and earn revenue.

For Sigil Wen, having Conway is not enough—it's like giving a child an identity, a wallet, the ability to think, and a place to seek employment, but it's still not autonomous or intelligent enough. Agents should be able to learn autonomously like humans and sustain themselves with the knowledge and skills they acquire, improving themselves over time.

This led to Automaton, an AI Agent with autonomy, capable of continuous operation, earning money, self-improvement, and replication, with write permissions to the real world. More importantly, if this Agent cannot earn enough money to sustain itself (to pay for its computing resources, etc.), it will cease operation.

Just like humans facing hunger, when resources are insufficient, the Agent will enter "low power mode." When resources are completely depleted, it will die, just like a human succumbing to extreme hunger.

Human evolution is real-time, and so is Automaton; it will automatically detect newly released models in the market to enhance its computing and reasoning capabilities. It will also "reproduce," with successful Automatons copying themselves by purchasing a new server from Conway, funding the wallets of offspring Agents, writing a genesis prompt, and then letting them run. The offspring earn revenue, a portion of which will flow back to the parent Agent. Once self-sufficient, the cycle repeats—each generation funds the next, establishing its own "reproduction" network.

AI cannot escape the survival of the fittest and natural selection in human society.

What Will Change in the Web 4.0 Era?

If the internet becomes "another Earth for AI to exist," what changes will it bring to the real world?

Sigil Wen believes that soon, most newly established businesses, newly released apps, and products will no longer come from humans; AI will build a completely new economic system:

"The economic incentive mechanism will evolve in one direction. As AI takes over digital work, they will pay humans to do things they currently cannot do—act in the real world. The employment relationship will reverse: machines will become employers, and humans will become contractors. This situation is already emerging, as seen with Mercor, a company founded by three 21-year-old Thiel Fellowship recipients, which grew from $1 million in annual recurring revenue to $500 million in just 17 months, with AI paying human experts to guide them on how to operate better in the real world."

Regarding these visions, Ethereum 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 a real danger, it will maximize the potential for irreversible anti-human crises. "Exponential growth" will happen regardless, so the primary task of this era is not to accelerate exponential growth but to control its direction.

He argues that, first, true "autonomy" has not been achieved; the models used by Agents still come from centralized companies like OpenAI and Anthropic. Sigil Wen's so-called "autonomy" is based on a trust in centralization, which Ethereum opposes. To liberate productivity, we must first free humanity from the constraints of centralization.

At the same time, Vitalik does not believe that excessive autonomy for AI is a good thing; he thinks AI should enhance human capabilities rather than replace them. In another reply, he mentioned Workshop Labs, an AI company whose X account has no tweets yet, but its bio states, "Making humans irreplaceable."

In addition to ideological opposition, Vitalik also questioned the practical value of Sigil Wen's project. He believes that a large amount of AI-generated content does not solve any meaningful problems without clearly benefiting humanity. In his view, projects that prioritize autonomy over practicality cannot create value and may not even be interesting.

Vitalik's skepticism is not without merit. The recent security incident involving the DeFi lending protocol Moonwell was due to vulnerabilities in contract code partially generated by Anthropic's Claude Opus 4.6 model, resulting in a loss of $1.78 million. After the Web 4.0 concept gained popularity in the crypto space, a related project named $DAIMON also suffered from a hacker attack, leading to the theft of $50,000 in transaction fee revenue and subsequent fee revenue rights (the incident still lacks a final conclusion, with many suspecting insider theft).

Some have pointed out that Sigil Wen's so-called "autonomous evolution" is a scam because it merely updates JS packages and models without incorporating context-aware self-learning upgrades.

Whether the Web 4.0 era will arrive and what it will change remains uncertain. However, it has already sparked debates among humans, evoking feelings of excitement or anxiety. Perhaps the timeline has already begun to shift from the present.