The only premium left is judgment

What will customers still pay for? The only premium left is judgment, and the ability to act on it before the client feels the need to ask.

Global AI spending in banking alone will reach USD 97 billion by 2027. The 2026 Cambridge report, covering 628 financial institutions across 151 countries, reveals something counterintuitive. The most widely adopted AI use cases are not the most economically valuable. Process automation, customer support and data visualisation dominate deployment across the industry. Yet the strongest profitability outcomes are reported in areas with far lower adoption rates.

• Treasury and ALM: 69% profitability for 28% adoption.
• Finance and FP&A: 62% for 37%.
• New product creation: 60% for 42%.

The pattern is not accidental. These are not execution use cases. They are decision-intensive use cases. Treasury models liquidity risk before it materialises. Finance and FP&A project future scenarios before capital is allocated. New product creation identifies emerging demand before competitors see it. The common thread is the same: AI improves the quality of decisions under uncertainty. That is where the premium lives.

A shock made it legible

The Middle East conflict made this legible at scale. Geopolitical shocks compressed the lead time that risk models had always assumed. Clients stopped rewarding speed. They started paying for the advisor who had already mapped their exposure before the conversation began.

JPMorgan read that shift early. In May 2025, the bank launched a Center for Geopolitics, USD 4.4 trillion in assets, advised by former senior defence officials. One mandate: translate geopolitical signals into client decisions before the market prices them in. HSBC, in the same window, dissolved its dedicated geopolitical risk team. Same environment, opposite conclusions about what clients would keep paying for.

Decision advantage as the source of value

Structured decision intelligence is exactly what AI now makes scalable across an entire book of clients. Reading signals before the call. Modelling exposure before the client formulates the question. The firms already doing this do not talk about AI strategy. They talk about client outcomes. The stack is invisible. The judgment is not.

Clients will continue to pay for the institution that understood their situation first, interpreted it correctly, and acted before they had to ask.

Cambridge Centre for Alternative Finance, 2026 Global AI in Financial Services Report, fig. 2.1, p. 34.