Why T+2 settlement is broken

Settlement in US public equities moved from T+5 to T+3 in 1995. T+3 to T+2 in 2017. T+2 to T+1 in May 2024. Three changes in thirty years. The chip in your phone iterates faster than the global settlement layer that clears most of the world's wealth.

I had a treasury lead at a fintech ask me last quarter why we don't just use blockchain to make T+0 happen. The honest answer takes longer than the question. Here it is.

Why T+1 is where it is

It's not that the technology is hard. The DTCC clears trillions a day. The bottleneck is that risk modeling is conservative, every shortened cycle requires a coordinated rule change across thousands of broker-dealers, and each step requires the entire industry to upgrade its operational tooling at once. T+1 took seven years of FINRA, SEC, and SIFMA work to get right.

What the float actually costs

When a customer trades $10M of equities at a US neobroker, that broker is on the hook for delivering shares, or cash, at T+1. Until then, capital is locked. Most neobrokers hold a multi-day reserve of float against this. Capital that earns the risk-free rate, not the firm's cost of capital. On a $5B AUM book, real money. PFOF, sweep yield, and similar mechanics exist in part to recover that gap, which I think is one of the under-discussed reasons retail-broker margins look the way they do.

What blockchain actually replaces

On-chain create-order finalizes on Base in under two seconds. That's the user-visible on-chain leg: stablecoin pulled, order_id minted, observable in the user's wallet immediately. The broker-to-exchange leg still settles T+1 (US equities) or T+2 (other markets) with the exchange, and the live Flo token mints via settle(order_id) only after that broker leg confirms. The exchange isn't a smart contract, and any vendor pretending the user-facing token mint or stablecoin payout happens before the broker cycle is selling you something.

What changes is the shape of what the user is waiting on. Without our infrastructure, every user redemption inherits an opaque T+1 wait with no on-chain artifact. With our infrastructure, the user has an on-chain order_id from the second the create-order tx confirms — observable, recoverable via user_cancel after the recovery window, immune to reorg/double-spend by construction. The broker cycle still gates the terminal token mint or stablecoin payout, but the wait has a contract-level receipt instead of a black-box ledger entry.

On-chain create-order doesn't eliminate T+1. It replaces an opaque off-chain wait with a contract-level claim the user can read, recover, and reason about.

What it doesn't replace

The broker-dealer is still required for asset segregation. The SEC isn't going to let a smart contract hold customer assets in good standing, and I don't think they should. The exchange is still required for price discovery. The DTCC is still required for clearing institutional flow. None of that goes away.

What does go away, from the user's vantage point, is the part the user actually sees. Token in a wallet. Available for transfer, lending, composability. Instantly. Everything before that, we wear. That's the trade.

T+0 will happen, eventually. Not because the underlying tech is the bottleneck (it isn't), but because regulators, broker-dealers, exchanges, and clearing houses move at the speed of policy, not engineering. Tokenization isn't waiting for that. The user already gets T+0 on the part they see. The legacy plumbing catches up on its own timeline. Both can be true.