When Should Boards Adjust Incentive Plans for AI Investment?

Artificial intelligence is becoming one of the largest capital allocation decisions facing corporate boards. Yet the financial measures commonly used in incentive plans may only provide a partial view of those investments, particularly when contractual commitments extend beyond the current performance period. This raises a question for compensation committees: How should incentive plans treat AI investments when their costs and benefits fall across different performance periods?

The starting presumption should be that management-controlled investment decisions remain reflected in incentive results, unless doing so would materially distort the performance bargain established at the beginning of the year. 

The question becomes more difficult when management accelerates AI investment beyond the original budget. Management may argue that executives should not be penalized for making a strategically important long-term investment. Yet shareholders bear both the cost and risk of that decision. Automatically excluding incremental AI investment from incentive results may unintentionally insulate executives from the cost of their decision, while shareholders remain exposed to whether the investment ultimately produces an adequate return. 

The Scale of AI Investment Raises the Stakes

A recent Wall Street Journal analysis estimates that nine major technology companies have approximately $3 trillion of AI contractual commitments, much of it associated with AI infrastructure. While these companies represent the extreme end of AI investment, the magnitude underscores the increasingly significant capital commitments companies are making to pursue their AI strategies.

For compensation committees, the challenge is determining how those investments should affect incentive-plan outcomes when actual investment differs materially from the assumptions underlying the original performance goals.

Unexpected Is Not the Same as Uncontrollable

Compensation committees have long adjusted incentive-plan results for unusual events that can distort the measurement of management performance. As discussed in a recent Pearl Meyer article, Incentive Plan Design in Early-Stage Growth Companies: Navigating Goal Adjustments During the Performance Period, adjustments are generally most defensible when circumstances are outside management’s control. AI investment can present a different situation because the incremental cost may result from a deliberate management decision. Management control is therefore relevant, but not necessarily determinative. The more fundamental question is whether including or excluding the cost provides a better measure of performance against the objectives the committee originally intended to incentivize.

There is a distinction between budget variance and performance adjustment. Companies routinely spend more or less than budget on strategic priorities. AI spending exceeding budget does not, by itself, justify an adjustment. The committee should instead ask whether changed circumstances have made the original performance goals an inappropriate measure of management performance.

Start With the Original Performance Bargain

When evaluating an adjustment, the committee should begin with what was known or reasonably foreseeable when goals were established, who controlled the subsequent decision to spend, and whether circumstances genuinely changed.

As AI becomes a strategic priority for more companies, committees should also be increasingly skeptical that significant AI investment was entirely unforeseeable.

Three Governance Factors to Consider

Start with economic substance. What has the company actually committed? Committees should look beyond the current-year accounting expense to capital expenditures, contractual obligations, incremental operating costs, and expected returns. 

Then consider shareholder alignment. Does the resulting incentive payout make sense in the context of shareholder returns, key financial outcomes such as cash flow and return on invested capital (ROIC), and any available evidence that the AI investment is creating measurable value? 

Finally, apply a symmetry test. Symmetry does not require mechanically identical treatment of overspending and underspending. Still, committees should understand whether their adjustment practices systematically protect management from unfavorable outcomes while allowing favorable outcomes to increase payouts. 

Consider the Time Horizon

Once the committee has evaluated whether an adjustment is warranted, it should consider both the current incentive period and the longer-term implications: 

• How should the cost be treated today?
• How should value be recognized over time?
• If significant AI investment becomes an ongoing part of the strategy, should future incentive plans be designed differently?

Could a Multi-Year Recognition Approach Be Appropriate?

Given its complexity, a multi-year recognition approach should generally be reserved for significant investments where current-period accounting treatment would materially distort the intended performance measurement. Accounting timing alone should not determine incentive treatment.

Strict current-year treatment can create a mismatch when a substantial investment is expected to generate benefits over several years. In those circumstances, committees may consider whether spreading the cost over a defined period provides a better measure of management performance than either recognizing the full cost immediately or excluding it altogether.

This approach can be reasonable, but it effectively creates a separate incentive-measurement framework that may differ from the company’s accounting treatment. To avoid becoming a disguised add-back, the committee would need clear and consistent rules for the recognition period, treatment of deferred amounts and impairments, and the impact on future management teams, while ensuring that costs deferred today are ultimately reflected in future incentive results.

Multi-year recognition addresses how the cost is reflected in incentive results, though it does not necessarily address how management should ultimately be rewarded for the value the investment creates.

Recognize Long-Term Value Through Long-Term Incentives

Management should not avoid an attractive long-term investment simply because it reduces the current year's bonus. At the same time, committees should be cautious about using year-end discretion to reward an investment before it demonstrates value. Long-term incentives provide a more appropriate mechanism for recognizing that value as it becomes measurable through total shareholder return, ROIC, cumulative cash flow, or other measures of long-term value creation. 

Address the Issue Prospectively

Finally, repeated requests to adjust for AI investment may signal that the underlying problem is actually the incentive design itself. If significant AI investment is expected to continue, committees should address it prospectively through goal setting, performance ranges, metric selection, or long-term incentive design rather than relying repeatedly on year-end discretion or bespoke multi-year adjustments.

The Bottom Line for Boards

There is no single answer for how AI investment should be reflected in incentive plans. Depending on the circumstances, committees may leave the full cost in the incentive calculation, use a carefully structured multi-year treatment, or address expected investment prospectively through incentive design. The right approach depends on the nature of the investment, the original performance goals, and the time horizon over which the costs and benefits are expected to occur.

AI is a particularly visible example of a broader incentive-governance challenge: how to align compensation with long-term value creation when an investment’s costs and benefits occur in different performance periods. The basic bargain should remain unchanged. Management should remain accountable for the capital it deploys and be rewarded when that capital creates value.