Upstart Wins Conditional OCC Approval for Branchless National Bank Charter
The Office of the Comptroller of the Currency has granted Upstart conditional approval to establish a branchless, full-service national bank, four months after the AI lending marketplace filed its application. Finovate reported the approval on 24 July. The proposed entity, Upstart Bank, would be headquartered in Delaware and operate without physical branches.
The charter Upstart is pursuing is a de novo national bank charter, a newly formed full-service bank rather than the acquisition of an existing one, and it brings with it the accompanying bank holding company oversight. Once fully approved, Upstart Bank would be able to originate loans to consumers in all fifty states and accept FDIC-insured deposits.
"Conditional approval from the OCC is an important milestone for Upstart Bank and we will continue to work with the OCC, the FDIC, and the Federal Reserve on the remaining steps," said Paul Gu, co-founder of Upstart. "Upstart Bank will allow us to lower the cost of lending and bring our full product offering to all 50 states, advancing our mission to radically reduce the cost and complexity of credit for all Americans."
The structural point is what the charter removes rather than what it adds. Upstart, founded in California in 2012, uses AI models to price credit and automate origination, but it has never held the licence that lets it lend in its own name nationally. It has instead relied on third-party bank partners to originate the loans its models underwrite, a rent-a-charter arrangement that carries a fee, a dependency and a supervisory relationship Upstart does not control. A national charter collapses that layer.
What it does not collapse is the funding model, and Upstart has been explicit that the bank is intended to complement rather than replace it. Banks, credit unions and institutional investors are expected to continue purchasing the vast majority of loans originated on the platform. That distinction matters: a lender that keeps its originations on its own balance sheet needs capital against them and takes the credit risk directly, whereas Upstart's economics have been built on selling the paper on. The company is buying origination independence, not converting itself into a balance-sheet lender.
The approval also lands in the middle of a visible reopening of the charter route for FinTechs, and it is worth being precise about how conditional the conditional approval is. The OCC's rejection of Wise's application demonstrates that a filing, even from a large and well-capitalised firm, is not a formality. Upstart still has remaining steps in front of the OCC, the FDIC and the Federal Reserve before Upstart Bank can open, and the conditions attached to a de novo approval typically cover capital, liquidity, business plan adherence and management before the charter is made final.
For the banks currently doing Upstart's originating, the read-across is uncomfortable. Partner-bank arrangements have been a reliable source of fee income for a small number of institutions willing to lend their charters to FinTech front-ends, and the model rests on the assumption that the FinTech cannot easily obtain a charter of its own. Every conditional approval erodes that assumption. The partner banks that retain the relationship will be the ones supplying something a charter does not confer, deposit funding at scale, balance-sheet capacity or distribution, rather than regulatory access alone.
The wider signal is about where AI-native credit underwriting ends up sitting in the regulatory perimeter. Upstart's pitch has always been that its models price risk more accurately than the FICO-anchored incumbents, a claim that is easier to make from outside the supervisory regime than inside it. Bringing that model under direct OCC, FDIC and Federal Reserve oversight subjects it to model risk management expectations, fair lending scrutiny and examination in a way that a bank partnership only partially did. If Upstart clears those conditions, it will have established that an AI-first lender can hold its own charter, which is a considerably more consequential precedent than the individual approval.
