The remittance story is the one everyone tells: a worker abroad sends money home, and stablecoins cut a 6% fee to something near zero. It is true, and it is a small fraction of what we actually observe.
The dominant flow in our corridor data is business-to-business working capital. An importer in Paris paying a supplier in Shenzhen is not solving a fee problem. They are solving a settlement-time problem and, more often, an access problem.
Median settlement in that corridor through correspondent banking was four business days when it worked. Through USDC on a low-fee chain it is under three minutes. The fee saving is real but secondary; the working-capital saving of not having four days of inventory in transit is an order of magnitude larger.
The second pattern is dollar savings rather than dollar payments. In markets with 30%+ annual currency depreciation, a stablecoin balance is a savings account, and the on-ramp fee is amortised over months rather than charged per transaction.
Both patterns imply a different product than the remittance framing does. They need deep local-currency liquidity at the edges, predictable on-ramp pricing, and business account features — not a cheaper Western Union.