It was 2:17 a.m. in Lagos. The naira had just dropped another 8% in the parallel market. A settlement batch from three corridors was stuck, because the correspondent bank had quietly de-risked the relationship two weeks earlier. We were moving float between nostro accounts by hand while the P&L bled from currency depreciation. That wasn't a scaling challenge. It was survival.
The reason to tell that story isn't drama. It's that the night was entirely predictable in hindsight, and nothing about it was a rail problem. The rails were fine. The treasury function wasn't.
If your treasury team is still managing float traps, reconciliation debt, and FX-timing gaps by hand in 2026 — while product and growth announce new rails and partnerships — you are not behind on infrastructure. You are behind on the part of the business that infrastructure can't fix for you.
Most African fintechs still treat treasury as a 2018 back-office function: prefund correspondent accounts, hope the rails clear on time, hope volatility doesn't wipe the month's margin. New rails — stablecoins, PAPSS, instant domestic systems — are necessary and not sufficient. The constraint has shifted from rail concentration to liquidity concentration. The teams that survive a bad week are the ones that stopped treating cash as one undifferentiated pool.
The Liquidity Ladder
I learned to layer cash the hard way, after watching an unlayered pool nearly cost a corridor. Three rungs:
Survival cash (bottom rung): 30–60 days of runway in the most liquid, least volatile form, plus immediate FX buffers for the corridors that can kill you fastest.
FX buffer (middle rung): corridor-specific liquidity for settlement gaps and volatility — the $150k–$350k a month that, on a bad week, used to sit trapped in transit.
Growth reserve (top rung): capital freed for new corridors and product bets — only after the first two rungs are solid.
Most teams live entirely on the bottom rung, with everything above it improvised under pressure. That's why a "positive FCF" announcement so often sits on top of economics that haven't been stress-tested. The ladder isn't sophisticated. It's just deliberate — and deliberate is the thing that's usually missing.
A self-audit
Five signs the treasury function is still operating like it's 2018:
"Treasury" is one person reconciling spreadsheets at month-end.
You prefund correspondent accounts days or weeks ahead with no real-time view of trapped float.
FX moves hit the P&L harder than they should, because there's no layered buffer.
New corridors or stablecoin pilots get announced before the liquidity and reconciliation architecture is stress-tested.
You depend on one or two banking partners for off-ramps in a key market.
Three or more, and the function is still 2018 — regardless of how modern the rails look on a slide.
What the night actually taught me
The fix that mattered wasn't a better rail or a better dashboard. It was the discipline to know, before the parallel market moved, which corridors could kill us fastest, how much buffer each one needed, and what the fallback was if a single counterparty walked. None of that is visible in revenue or volume. All of it is visible at 2:17 a.m., when the system is under stress and the layering either holds or it doesn't.
The next 24 months won't be won by whoever announces the most partnerships. They'll be won by the teams that actually control liquidity, settlement timing, and their regulatory standing in the corridors that matter — quietly, before the night that tests it.
ACSS Corridor Intelligence — corridor risk, settlement architecture, and treasury strategy across African cross-border payments. New issues at ci.acss.africa. If your treasury is firefighting a corridor right now, reply and tell me which one — that's usually where the real architecture work begins.