OakNorth, a UK digital bank that has adopted an artificial intelligence (AI)-based credit rating system, is drawing attention for improving its earnings and soundness indicators while simultaneously cutting costs. Amid growing debate over overhauling Korea's credit evaluation framework following the presidential office's criticism of "cruel finance," observers say Korea also needs to apply more sophisticated AI models.
According to Toss Insight's "Monthly Insight" report released Wednesday, OakNorth posted net profit of 165.2 million pounds (about 330.1 billion won) last year, up 3.7% from the previous year.
New loan originations reached 28 billion pounds last year, a 33% increase from the previous year. Despite the expansion, the cost-to-income ratio (CIR) fell 3 percentage points from 29% to 26%.
Founded in 2015, OakNorth is a UK digital specialty bank. It has grown by primarily targeting high-growth small and mid-sized enterprises that struggle to clear traditional banks' thresholds due to insufficient collateral or credit ratings.
Over the decade since its founding, the bank has supplied more than 15.1 billion pounds in loans to small and mid-sized enterprises while recording a cumulative principal loss rate of 0.045%, managing soundness at a virtually loss-free level. Loan balances grew 18% from 6.1 billion pounds to 7.2 billion pounds last year, and pre-tax profit reached 222.5 million pounds.
Behind this performance is the AI-based credit analysis platform ONCI. While traditional screening relies on past financial data and captures shifts in borrower risk with a lag, ONCI analyzes forward-looking data, including industry-specific scenarios, sectoral impacts from AI adoption, and climate risks. Based on this analysis, the platform calculates default probabilities and forward-looking ratings (FLR) for each borrower, improving screening accuracy. This enabled the bank to expand lending without increasing headcount, minimizing costs.
OakNorth signed a strategic partnership with OpenAI in May last year. Deploying multiple AI agents, including ChatGPT and Claude, to boost operational efficiency also helped improve the CIR. Expanding lending in the relatively faster-growing US market also contributed to its performance.
"Through AI adoption, OakNorth increased new lending by 33% year-on-year last year and achieved a CIR of 26%," said Noh Kyung-ah, a research fellow at Toss Insight. "This suggests that introducing an AI-based credit evaluation system can help financial institutions expand operating leverage and improve their cost structure."
