Three Questions Your Board Should Be Asking About AI Spend

Boards are good at asking the right questions about capital allocation. They do it for property, for headcount, for technology infrastructure and for M&A. They have the frameworks, the vocabulary and the institutional muscle memory to hold management accountable for returns.

For AI spend, that muscle hasn’t been built yet.

Not because boards are avoiding the conversation, most boards are actively discussing AI. But there’s a difference between discussing AI as a strategic opportunity and exercising genuine financial oversight of AI as a significant and growing capital allocation. Most boards are doing the former. Very few are doing the latter.

That gap matters. Token spend is scaling rapidly and the organisations that do not build board-level accountability frameworks now will be having a harder conversation in twelve to eighteen months.

The question boards are asking and why it isn’t enough

“Are we using AI?” Yes. “Are we keeping up with competitors?” Probably. “What is our AI strategy?” We have one.

These are strategy questions, not governance questions. They tell you whether the organisation is moving in the right direction. They don’t tell you whether the investment is being managed responsibly.

Question one: Where is the AI budget going — specifically?

Not the total. The breakdown. Which teams are consuming inference spend? Which workflows? Which vendors? And when spend is attributed to a business unit, what outcome is it producing?

If your management team’s answer is a consolidated invoice total and a set of qualitative examples, the governance framework is not yet in place. Token spend that cannot be attributed at the workflow level cannot be managed, rightsized, or held accountable.

The board’s role is not to demand technical granularity. It is to require that a shared framework exists between Finance and Engineering for attributing inference costs to business outcomes and to hold management to building it.

Question two: What is the AI investment producing — measurably?

“We believe AI is adding value” is not an answer a board should accept for a material capital allocation.

Enterprise AI deployments face 20 - 40% margin compression without governance in place. The organisations managing AI investment well have moved past narrative to numbers: cost per outcome for key AI workflows, efficiency gains expressible in time or resource terms, revenue impact attributable to AI-enabled processes.

cortave reduces token spend by 50 - 80%, making real-world AI deployments financially viable. That kind of return is measurable. If your current AI investment can’t be measured the same way, that’s the signal.

Question three: Do we know what AI is being used that we haven’t approved?

This is the question most boards are not asking. It may be the most consequential one.

Token spend from unapproved, untracked AI usage such as teams running model integrations and API connections without IT or Finance oversight is growing in every organisation that has adopted AI at scale. An organisation that has approved $200,000 of AI spend and has no visibility of an additional $60,000 in unapproved usage is not managing its AI investment. It is managing the part it can see.

Requiring a token spend audit is a reasonable and proportionate board-level ask. The organisations that have done it have consistently found the unmanaged estate is larger than expected.

What governance-ready looks like

Boards exercising effective AI oversight share one characteristic: they have asked management to build the infrastructure that makes accountability possible. Not to slow AI adoption but to ensure that as token spend scales, the organisation can demonstrate returns with data rather than narrative.

That infrastructure has four components: visibility at the workflow level, attribution of inference spend to outcome, guardrails that prevent token waste without blocking innovation and a reporting framework that gives the board a clear summary.

We’ve created a Board AI Governance Briefing Pack to give your board the frameworks, metrics and oversight structures needed to govern AI investment. Download it here