TradFi Wants Blockchain, Not DeFi

Not convergence, extraction

The canonical crypto story holds that DeFi and traditional finance are converging into a single permissionless system where the new subsumes the old. a16z's crypto research team calls that story "mostly wrong." The honest version: "where TradFi can use a blockchain to make its existing business better it will. Not because it has embraced decentralization, but because it's a compelling COGS story," meaning the technology happens to cut cost, improve settlement, expand distribution, and tighten a firm's grip on its own customer relationships.¹

Two filters together govern what actually gets adopted. First, a cost-of-goods test: does the primitive improve cost, risk, or distribution. Atomic settlement collapses the gap between a trade and its finality, erasing counterparty risk and freeing up collateral that would otherwise sit parked; tokenized money-market funds from firms like BlackRock and Franklin Templeton put yield-bearing assets on-chain; stablecoins move dollars programmatically.¹ Second, a control test: the primitive must not require surrendering institutional control. "If a feature delivers value only by removing institutional control, it will almost certainly be reshaped or rejected, however elegant it may be," which is why pseudonymity, fully permissionless composability, and trust-minimization as an ideology all get discarded even as the underlying rail gets adopted. JPMorgan runs a permissioned deposit chain, Circle ships Arc, an institutional product built for compliance and known counterparties rather than open access, and Apollo's ACRED wraps DeFi primitives like Morpho in guardrails rather than adopting them as-is.¹ A corollary follows directly from the control test: the best technology does not automatically win inside an enterprise, since institutions buy what fits existing workflows, risk models, and procurement processes, not what is theoretically superior.

Why it matters

This is the permissioned pole of a broader two-lane view of crypto adoption, and it functions as a design brief for anyone building on-chain infrastructure aimed at institutions: expose atomic settlement, tokenized collateral, and programmable dollars, and do not require counterparties to abandon compliance or control in order to get them.

CZ's counter-prediction

CZ forecasts a genuinely different end state: full convergence into one industry rather than a stable third category. "There shouldn't be a crypto finance versus what we call mainstream or traditional finance...you don't have postal mail versus email...there's not two parallel financial industries. It should just be one."²

His evidence is the same evidence behind the institutional-adoption thesis, stocks being tokenized, banks using blockchains, but read as a merger already in progress rather than as institutions extracting the useful rails while discarding the underlying ideology. The disagreement is real rather than semantic: the institutional-pragmatism view predicts programmable financial infrastructure as a durable third thing, permanently optimized for compliance and control, while CZ predicts the crypto and traditional finance distinction simply dissolves, leaving one industry behind. His own analogy arguably concedes part of the opposing case without meaning to, since email did not merge with postal mail, it replaced it for most practical purposes, which is closer to extraction than to convergence. There is also a structural interest worth noting: an exchange founder benefits from a framing in which crypto is not a sector but the substrate.²