Artificial intelligence promises to lift bank profitability and reshape competitive rankings, but the path is proving far slower and more hazardous than early enthusiasm suggested. S&P Global estimates that rated banks’ average return on equity could climb from 12% to 14% over the next three to five years—yet nearly half of all AI initiatives within banks currently fail, according to the agency’s own researchers. The industry is caught between the urgency to adopt and the dangers of moving too fast.
In brief
- —Over 90% of banks are already engaged in AI adoption
- —Nearly half of bank AI initiatives fail before delivering returns
- —Three US banks hold 75% of all industry AI patents
From 12% to 14% ROE: the reward banks are chasing
The financial case for AI in banking is clear enough on paper. S&P Global‘s Madrid-based lead researcher on AI adoption, Miriam Fernandez, puts the prize in concrete terms: rated banks’ average return on equity rising two percentage points over the next three to five years, driven by automation and efficiency gains across the industry.
