Pull up your 2027 budget draft and search for “AI.” You will probably find it buried as a line item inside IT operations. At some credit unions the CFO filed it under “digital banking enhancements” because that category seemed close enough. At others the finance team parked it in professional services as a catch-all. In each case, somebody else owns the parent category, so AI spending competes with servers, core conversions, and branch renovations for the same dollars, reviewed by a person whose day job is those other things.

Budget season is when you fix that. AI in 2027 should be its own category, reported to the board the way you report marketing or facilities. The reason is practical: its cost structure, decision pace, and organizational reach cut across too many departments to sit neatly inside any one of them. Leaving it buried distorts both the AI work and the category it is hiding in.

Try the look-back test first

Before anyone in your budget meeting anchors to specific vendor names or product prices, run this exercise. Ask your CFO to think back to September 2025. Could they have predicted which AI tools your team would be seriously evaluating right now? Could they have scoped the use cases that ended up mattering?

We have asked this at every budget planning session we have joined this year. Nobody said yes. The tools that look obvious in fall 2026 either did not exist or were not production-ready twelve months ago. Use cases that seemed speculative in 2025 became routine by summer 2026. And the pace keeps accelerating. We think 2026’s AI progress will look modest compared to what 2027 delivers.

The practical upshot: if you could not have predicted where the end of 2026 landed from where you stood last September, you cannot line-item the end of 2027 either. A line-item budget assumes you know what you are buying. A category budget says you know what you are committing to, without forcing the CFO to pretend anyone can see twelve months out in a space moving this fast.

Why not just keep it inside IT

When AI costs cross every department, housing them inside IT creates problems that compound as the year goes on. A back-office automation pilot, for instance, touches operations staffing, vendor contracts, compliance documentation, and training all at once. A member-facing chatbot ripples into marketing, contact center operations, and your exam file. Parking all of that under the CIO makes the CIO a bottleneck for spending decisions that really need the COO, CCO, and CEO in the room. It also frames AI as technology overhead, which is the fastest way to get it cut when somebody needs to find a few basis points in Q3.

Credit unions that frame AI as a growth investment, the way they frame marketing or a new branch strategy, ask a fundamentally different budget question. Instead of “which specific AI tools are we buying,” the question becomes: what are we willing to commit this year to keep pace with the institutions we compete against and differentiate the member experience enough to matter? That is a commitment question. Commitment questions belong at the category level, not buried inside someone else’s operating budget.

Sizing it without line items

Any CFO will push back here, and they should. A budget category with no detail underneath it is uncomfortable. Boards approve categories expecting line items. Here is how to handle the tension without papering over it.

The conversation starts with a percentage-of-operating-revenue figure at the executive level. Your board already thinks in these terms for other categories, so the framing is familiar. The specific number varies by institution and by appetite. What matters is that leadership picks it on purpose rather than backing into a figure from a single vendor quote that someone emailed the CTO in August.

Then split the category into quarterly allocations instead of locking everything into an annual plan. AI moves in quarters. A twelve-month plan signed off in October will feel dated by March, and if the last two years are any guide, real opportunities will surface that nobody scoped during budget season. Quarterly reallocation lets the team redirect dollars when that happens instead of going back to the board for a mid-year amendment.

Assign one executive owner for the whole category. The right person is whoever drives operational strategy across the institution, whether that is the CIO, the COO, or a chief digital officer. The owner approves quarterly spending, reports deployment outcomes to the board, coordinates with compliance on the exam file, and connects the three-signals screening work to spending authority. This is also the person examiners will ask for by name, so the choice should be someone who can speak to both the business case and the risk controls.

If your team already ran a 90-day pilot, you have real cost data to anchor Q1. If not, Q1 spend is the pilot itself, and the remaining quarters flex based on what you learn.

Examiners are already asking

Credit unions sometimes avoid creating a visible AI budget category because they assume it draws examiner attention. That concern is backwards. NCUA does not have AI-specific regulation today, but examiners are already reviewing AI deployments under existing third-party vendor, information security, and internal controls frameworks. The exam questions we documented in August show the kinds of questions field examiners are posing right now. A clean budget category with documented governance, quarterly reviews, and a named owner gives you a straightforward story to tell in an exam. Compare that to the alternative: AI expenses scattered across six departments where nobody can produce a consolidated view when the examiner asks for a summary of your AI spend and governance in one place.

What you risk by leaving it buried

The real cost of keeping AI as a line item inside IT is not overspending. It is spending the right amount in the wrong way: locked to a vendor you chose in October for a problem that shifted by April, governed by a department that treats AI as one more infrastructure line, invisible to the board members who should be asking whether the institution is moving fast enough.

Making AI its own category will not make every dollar land perfectly. No budget structure does. But it puts spending where leadership can see it, govern it at the right level, and treat it as what it is: a multi-year bet on whether your credit union can compete.

Budget season closes once. Get the structure right before the numbers go to the board.

If your team is heading into 2027 budget conversations and wants an outside perspective on where AI fits, Advisor Labs runs a 45-minute back-office AI audit that covers current-state assessment, category sizing, and governance structure. Book the conversation here.

For weekly analysis like this delivered to your inbox, subscribe to the AiForCU newsletter.