Bbva bets the bank on ai while war and energy shocks rattle the boardroom
Carlos Torres Vila didn’t wait for the applause to finish before dropping the next pivot: BBVA will mint its next decade of profits from lines of code, not collateral. Speaking to shareholders in Bilbao on Friday, the chairman declared artificial intelligence “the great transformer of our time” and revealed that every euro of the 2025-2029 strategic plan is now indexed to algorithms that most regulators haven’t even named yet.
Torres and genç tag-team the future
It was a choreographed one-two. After Torres painted the macro picture—war, gas prices, tokenized assets—CEO Onur Genç supplied the slide deck: 127,000 employees will soon be “augmented” rather than replaced, 11.5 million new customers arrived last year, two-thirds through mobile screens, and the €10.5 billion net profit for 2025 is already being recycled into GPU clusters and data-privacy firewalls. The bank, Genç admitted, still doesn’t know exactly what the models will look like in 2027. “But we know speed will be the only moat left.”
The immediacy is what rattled analysts. While rivals flirt with chatbots, BBVA has quietly split its AI war chest into three live workstreams: client-experience hyper-personalization, back-office shrinkage, and trader co-pilot tools that flag rogue positions before humans smell smoke. Internal dashboards seen by TechBloom show pilot credit-decision models cutting approval times from 45 minutes to 23 seconds with a 12 % drop in default probability—numbers the bank hasn’t published but whispered to investors over coffee between votes.
Record profits fund the gamble
The balance sheet gives BBVA room to move fast. Return on tangible equity hit 19.3 % last year, double the European peer median, and the cost-to-income ratio fell below 40 % for the first time in two decades. That efficiency engine is what lets Torres promise €3 billion in fresh tech capex without flinching on dividend policy. The catch: every basis-point rise in model risk-weighted assets now flows directly into the AI audit queue; regulators in Madrid and Frankfurt are demanding explainability logs in real time.
Outside the auditorium, staff describe a culture shift that feels more startup than staid lender. Cross-functional “squads” of data scientists, lawyers, and mortgage clerks sprint in six-week cycles, racing to ship features before the legal framework solidifies. One engineer joked that the compliance team has learned Python faster than the coders learned Basel III.
Still, the unknowns loom. Sam Altman’s prediction of superintelligence by 2028—invoked by Torres from the podium—hangs over the room like ozone after a lightning strike. If the timeline holds, BBVA will be negotiating loan covenants with entities that can read every footnote in the contract before the ink dries. The bank’s answer is a governance sandbox already running inside a segregated cloud, stress-testing future agent-to-agent negotiations on synthetic mortgage portfolios.
The shareholders approved the plan unanimously, but the real verdict will come from depositors who never read a 10-K. They will notice only whether the app guesses their life event before it happens and whether the interest rate offered feels like a favor or a heist. BBVA is wagering that the same code can do both. The next four years will tell if that confidence is vision or hubris—no footnotes required.