Joseph Lubin, co-founder of Ethereum and chief executive of software firm Consensys, has issued a pointed warning about the convergence of artificial intelligence and crypto, arguing that if AI infrastructure stays in the hands of a small number of technology giants, the decentralised economy will be in serious trouble.
Speaking ahead of an upcoming industry event, Lubin said autonomous and semi-autonomous AI agents are on course to become a central feature of blockchain ecosystems, handling transactions, verifying data and coordinating with other agents across decentralised networks.
"If AI infrastructure remains concentrated among large technology firms, we could be in trouble," Lubin said, framing the risk as a direct contradiction of the principles crypto was built on.
His argument rests on how users will interact with blockchains in the coming years. Rather than manually handling wallets, approving transactions and navigating smart contracts, Lubin sees AI systems becoming an intermediary layer — users will give instructions, and AI agents will execute them on-chain in the background. That, in his view, is what will finally make crypto genuinely accessible to mainstream users.
The flipside is dependence. If the AI systems handling those instructions are controlled by a handful of centralised companies, Lubin argues, the resulting infrastructure would reduce transparency, limit competition and create systemic risks — the very problems blockchain was originally designed to address.
Lubin pointed to Consensys's own MetaMask wallet as an example of how that convergence is already playing out. "You can walk around with your personal financial system in your pocket," he said, describing MetaMask's evolution from a simple wallet into what he calls a "personal financial operating system." In that model, users retain custody of their assets while AI agents manage routine transactions and strategies.
The Ethereum co-founder believes blockchain and cryptography can serve as a check on AI concentration. By operating in transparent, cryptographically verifiable environments, AI agents would be forced to prove their actions rather than rely on trust in centralised intermediaries. "AI agents wouldn't blindly trust centralised platforms," Lubin said, describing a world in which decentralised networks validate each other's actions.
Beyond the AI thesis, Lubin used the appearance to outline broader shifts he expects to see in Ethereum itself. He predicts the rise of "corporate chains," where companies build their own blockchain infrastructure for performance and control, but stressed that critical assets should still settle on Ethereum's main network to preserve long-term security. He also characterised stablecoins as merely an early stage in crypto's evolution, with more advanced decentralised financial systems likely to emerge.
The comments land at a moment when traditional finance is flooding into digital assets — Charles Schwab launched spot bitcoin and ether trading this week, Morgan Stanley has rolled out a spot bitcoin ETF, and Goldman Sachs has filed for its own bitcoin income ETF. For Lubin, that wave is significant, but the AI question is a separate and potentially larger battle over who controls the plumbing.
His bottom line is cautionary rather than alarmist. AI could unlock the next major cycle of blockchain adoption, he argues, but only if control of the underlying infrastructure does not end up concentrated in the hands of a few.