Global AI capital expenditure is expected to reach $740 billion in 2026, according to Morgan Stanley. Goldman Sachs projects AI companies alone will invest more than $500 billion. These are infrastructure numbers. They belong in capital planning, not discretionary technology budgets. Yet most financial services firms still account for AI spend as operational overhead, buried in technology run-costs or innovation budgets that lack the rigour of a formal capital case.
The boards I advise are losing patience with this.
The Overhead Trap
The pattern is familiar. An AI initiative starts as a proof of concept. It gets funded from a departmental innovation pot or a central technology budget. It succeeds. It scales. It begins consuming meaningful compute, data engineering time, and vendor fees. And it sits in the P&L as an operating expense, invisible to the capital allocation process that governs every other strategic investment the firm makes.
This matters because capital allocation is how boards express strategy. A firm that treats AI as overhead is telling its board, its investors, and its regulators that AI is a cost of doing business, not a source of competitive advantage. In 2026, that signal is wrong, and investors know it.
IR Impact's analysis of recent earnings calls found that capital allocation has become the equity story in the AI era. The lens through which institutional investors judge management teams is no longer just earnings growth; it is how intelligently and transparently a firm invests in AI capabilities. Financial services firms that cannot articulate their AI capital strategy in the same language they use for branch networks, trading platforms, or core banking replacements are leaving credibility on the table.
What the Leaders Are Doing Differently
PwC's 2026 AI Performance Study, surveying 1,217 senior executives globally, found that 74% of AI's economic value is being captured by just 20% of organisations. The distinguishing factor is not spend. It is how that spend is governed.
The top performers share three characteristics. First, they treat AI investment as a portfolio with explicit risk-return profiles, including a built-in failure rate assumption of 40 to 50% across initiatives. This prevents the common mistake of betting everything on one or two high-profile use cases and abandoning AI entirely after a single failure.
Second, they use AI for growth, not just efficiency. PwC's data shows that pursuing new revenue opportunities from industry convergence is the single strongest factor influencing AI-driven financial performance, ahead of cost reduction alone. The firms generating returns from AI are not automating existing processes more cheaply. They are building new revenue lines that did not exist before.
Third, they govern AI investment through the same capital allocation discipline they apply to every other strategic programme. That means a business case with quantified outcomes, a staged funding model with gates, clear ownership at senior management level, and reporting to the board through the same channels that govern traditional capex.
The CFO's Role
For financial services CFOs, the transition from AI-as-overhead to AI-as-capital-line-item requires three shifts.
The first is visibility. Most CFOs I work with cannot tell you, to within 20%, what their firm spends on AI across all business units. The fragmentation of AI spend across departmental budgets, vendor contracts, cloud compute, and internal labour makes consolidated reporting difficult. But without that visibility, capital allocation is impossible. The starting point is a consolidated view of AI spend, mapped to business outcomes, presented in the same format as every other investment category.
The second is appraisal. AI investments need the same rigour as any other capital commitment: a business case with assumptions, a measurement framework, and defined criteria for continuing, scaling, or exiting. The firms still running AI as "innovation" without formal return expectations are the ones stuck in pilot purgatory. Gartner's 2025 AI Maturity Curve found that only 11% of financial firms report measurable ROI from AI initiatives. The rest are spending without a capital discipline that would force the question.
The third is board reporting. AI should appear as a standing item in the board's capital allocation discussion, not as a technology update in the CTO's quarterly review. The audit committee needs to see AI spend alongside other material investments, with the same governance expectations. When three in four boards have approved major AI investments but fewer than half have set governance expectations or made AI risk a standing agenda item, the gap between commitment and oversight is a governance risk in itself.
What This Means for 2026 Planning
The financial services firms that will outperform in the next cycle are the ones making a deliberate choice now: AI moves from the innovation budget to the capital plan, with all the discipline that implies.
This does not mean every AI initiative needs a five-year NPV. Some experiments should remain small, fast, and disposable. But the portfolio as a whole needs a capital framework. How much is the firm investing in AI this year? What is the expected return profile, by initiative and in aggregate? What is the failure budget? Who owns the portfolio, and how is performance reported to the board?
Fortune's survey of CFOs for 2026 found a decisive shift in framing: finance leaders now predict AI transformation, not just efficiency gains. That prediction only converts to value if the capital allocation machinery is built to support it.
The firms still treating AI as overhead in their 2026 budgets are not just under-investing. They are under-governing. And in a regulated industry, that distinction matters more than the spend itself.
*To discuss how the 90-Day AI Acceleration programme can help your organisation build a board-ready AI capital case, contact the Value Institute.*