Vitalik Buterin calls AGI a big unhedged gamble on human relevance

AGI risk, according to the man who built Ethereum: “There’s a big gamble we’re taking on the future.”

The bet nobody signed up for

In a thread posted July 20, 2026, Vitalik Buterin laid out his concern: we are all betting that some range of human capabilities will stubbornly resist AI. He is not certain that range exists.

The fork in his scenario is stark. If that resistant range does exist, the next AI era looks amazing: every foot soldier becomes a manager, every manager becomes a general. Humans move up the stack as AI absorbs the work below them. If it does not exist, the era of guaranteed human relevance ends, and there is no rung left to climb to.

Power before intelligence

Buterin’s sharper warning is about power, not intelligence. He argues the more immediate danger is a small cabal of corporations or governments seizing unilateral control of transformative AI. Before the question of whether machines outgrow humans comes the question of which humans hold the machines.

His proposed hedge follows from that: deep human-machine integration through brain-computer interfaces and decentralized systems, so that “human plus machine” stays stronger than machine alone. It is a characteristically Buterin answer, distributing capability rather than concentrating it, applied this time to cognition itself.

A portfolio manager’s framing

What makes the thread land is its register. Coming from crypto’s most influential builder, this is not a doomer take. It reads like a portfolio manager describing an unhedged position, except the position is humanity, and the hedge he proposes (integration plus decentralization) is still mostly unbuilt.

The thread leaves one question hanging for everyone who reads it: is there a human skill AI will never absorb? Buterin’s honest answer, and the unsettling part, is that nobody actually knows.