Crypto's speculative frenzy and metaverse dreams didn't survive. What did was the financial plumbing. Yat Siu, chairman of Animoca Brands, now sees that infrastructure as something far more powerful than anyone originally pitched: the backbone for autonomous software agents to transact without human intermediaries.

The shift wasn't intentional. Looking back at gaming NPCs in Animoca's portfolio, Siu spotted the pattern only in retrospect. When OpenAI's model demonstrated what autonomous agents could do, and Animoca's backing of Ethoswarm's technology proved similar capabilities existed, the thesis clicked into place. CryptoSlam, which Animoca led a $9 million funding round for in 2022, was marketed as a blockchain data platform. The agentic angle emerged later, reframing what was already built.

The Infrastructure That Survived

Crypto shed its casino layer fast. The memecoins, the wallet friction, the "metaverse as place" gospel that crumbled, none of that mattered. What persisted was what Siu calls "machine banking." A digital wallet becomes a bank account for software. Stablecoins function as currency. The blockchain handles settlement. The whole stack was designed for machines from the ground up. Humans just needed to prove it worked at scale.

The practical model is cleaner than it sounds. Agents with wallets paying each other for skills, services, compute, and data through native tokens. No bank account required. No KYC forms. No human authorization delays. Machine-to-machine commerce that couldn't exist in traditional finance because no infrastructure existed for it. Siu frames it as an expansion, not a replacement. Human transactions keep running. A new surface emerges on top.

What the Industry Is Grappling With

The leap from theory to deployment has sparked divergent reactions. Some builders treat autonomous agents as inevitable infrastructure. Others remain skeptical that software wallets and token settlements solve real coordination problems between machines at scale. The technical track record of early systems matters less than the structural question: will sovereign money for agents actually create network effects, or does centralized cloud computing remain simpler for most use cases?

Siu's parallel to early online payments is telling. Thirty years ago, digital transactions seemed unnecessary when wire transfers worked. They weren't necessary until they became native to how commerce operated. The agentic economy isn't here because crypto was essential. It's here because the infrastructure, inherited from years of building against friction, finally met a problem that infrastructure solves better than alternatives.

This article is informational. It does not constitute financial advice or investment guidance.