Banks are yet to identify the right AI stack they need: the challenge for banks will be to tap the right talent for AI.

In recent years there is a lot more “parity” that public sector banks have achieved through digital technology, with private sector banks.
“Digital adoption is accelerating, with PSBs closing the gap with private banks in core capabilities such as digital onboarding, card and loan management, and AI support,” Peeyush Dalmia, senior partner at McKinsey & Company India said, in a banking report released on Tuesday.
“However, while private banks still lead on digital investment, allocating up to 10% of operational expenses to IT compared with 6 to 8% just a few years ago, public sector banks (PSB) are making progress,” Dalmia said.
Income per employee has also improved for PSBs: doubling to ₹0.8 crore in 2025 from ₹0.40 crore in 2019. In the same period, the income per employee for private sector banks has risen slower to ₹0.67 crore from ₹0.49 crore. But private sector lenders have a higher employee base in this period while PSBs saw the employee base get leaner due to the massive consolidation in 2020.
The introduction of digital tools, improved processes to boosting operational efficiency and strategic initiatives have been key to these improvements, the McKinsey report says.
The main difference remains in customer experience—the top three PSB apps score between 4.1 and 4.3, while the top three private banks score around 4.5—highlighting the need to improve app stability, design, and engagement to match private banks and boost customer satisfaction, the report says.
"But at least eight out of every 10 banks are still trying to figure out what their AI stack should look like. The system is yet to be institutionalised,” Dalmia told media persons, after releasing the report. This is in stark contrast to a non-banking financial company such as Bajaj Finance which has declared FY27 as its do-or-die year for AI transformation; where it is being treated as a once-in-a generation reset.
Bankers are finding out what the pockets of value are; how to save costs and how cyberfraud security can be strengthened.
AI on a customer facing application, at this stage, in a regulated industry, will be very difficult to be all pervasive, the report says. The value from AI is in three buckets: to deal with regulatory, compliance and reporting processes; the second is making processes more efficient so that value delivery to customer improves. The third would be on improving value to the organisation, in the form of sales productivity.
Dalmia also said that banks would continue to chase talent to improve their AI processes: these talent pools would be digital specialists, designers and engineers.
As a broader trend, Indian banks reached record ROAs in fiscal year 2025. “The profitability peak masks a deepening structural challenge—compressing NIMs, declining fee income, and high operating costs, which may affect future earnings potentials,” the report said.
While asset quality improved to a 13-year low, gross nonperforming assets (GNPA) ratio of 2.2% in fiscal 2025, elevated write-offs in unsecured retail and increasing slippages signal emerging credit stress that could affect future returns.