Vertically Integrated Programmable Bank
$112M/year Hereticon of nationally chartered banks
The key strategic move was owning the bank and the API at the same time, because it lets Erebor make money from both sides of the same customer relationship. A defense or AI startup can keep millions in deposits at Erebor, while a separate software team can plug into Erebor to open accounts, move ACH and wires, or build stablecoin payment flows. That is closer to Column’s vertically integrated model than to Mercury’s original software layer on top of partner banks.
-
Column shows what this model looks like in practice. It owns a bank charter, payment rails, ledger, and compliance stack, then sells direct API access for accounts, wires, cards, and money movement. That removes the usual sponsor bank middleman and turns infrastructure fees plus deposit spread into one combined revenue engine.
-
Mercury started from the other direction. It built the customer facing startup banking product first, then relied on partners like Evolve, Synapse, and later Column and Choice. Its later charter application shows how valuable direct control of the banking layer became after partner risk and migration pain surfaced.
-
For Erebor, this matters more because its customers are concentrated, cash rich, and operationally complex. A nuclear startup or stablecoin payments company is not just parking cash, it also needs custom treasury flows, account controls, and programmable money movement. Serving both use cases from day 1 raises revenue per customer and deepens lock in.
This points toward a new kind of startup bank, where the winning institution is not just the safest place to hold deposits, but the easiest regulated system to build on. If Erebor keeps pairing large balance commercial accounts with high value infrastructure customers, it can compound faster than a pure neobank and with more upside than a bank that only sells balance sheet products.