Insights
Boards and CEOs increasingly disagree about the pace of AI transformation. Investment intent is shared across both groups. Pace is where the disagreement concentrates.
Recent research quantifies this pattern. BCG’s Split Decisions survey, covering 351 CEOs and 274 board members at companies with at least $100 million in annual revenue, found that sixty-one percent of CEOs say their boards are rushing AI transformation.
A board sets the pace it believes the situation requires. The CEO carries the pace the organisation can actually execute. A gap this large between those two positions indicates a structural condition. It shows where the organisation’s understanding of AI actually sits. It also shows how far that understanding is from the timeline built on top of it.
A board sets its expected pace based on what it believes the organisation can do. When that belief is inaccurate, the pace becomes inaccurate as well. The CEO then absorbs the resulting gap.
This is the position 61% of CEOs describe. They carry a timeline set at a level of the organisation that does not hold the execution detail. The pace itself is a downstream output. It follows from an earlier judgment made with incomplete information.
Adjusting the pace without examining that judgment leaves the underlying constraint in place. The same mismatch reappears at the next initiative.
Confidence exceeds understanding. Seventy-five percent of board members describe their own AI literacy as sufficient or advanced. Forty percent of CEOs say their boards do not hold an informed view of what AI actually changes in the business. A board confident in its own understanding has little internal reason to slow down and verify it.
External narrative fills the gap. More than half of CEOs say AI hype is distorting how their boards evaluate AI decisions. A third say boards overestimate how much of the workforce AI can realistically replace. Where operational detail is thin, market narrative supplies the missing view. That narrative tends to underprice the difficulty of execution.
Accountability is distributed unevenly. CEOs estimate that 35% of their own performance evaluation depends on delivering AI returns. Boards estimate 27%. An eight-point gap in a single, quantifiable expectation is informative. It indicates that the party carrying the consequence is working from different assumptions than the party setting the pace.
A pace set at board level becomes a delivery timeline at the operating level. The mismatch therefore extends beyond governance into execution.
Separate industry data on AI value capture points to a similar shortfall. Around 60% of companies report little or no material financial benefit from AI investment so far. Only 5% show sustained profit-and-loss impact.
Read together, these two findings describe a connected dynamic. A board under-informed about execution difficulty sets an aggressive pace. The CEO absorbs that pace without resetting the board’s expectations. The organisation is then asked to deliver against a timeline built without full knowledge of what delivery requires. The shortfall surfaces later, in an AI initiative that does not reach its expected return.
Three implications follow from this pattern.
The first concerns the knowledge gap itself. Closing it depends on the CEO. A board that considers its own AI literacy sufficient has no internal signal telling it otherwise. That signal has to come from the CEO, in the form of direct detail about what the organisation’s data, processes, and people can currently support.
The second concerns the reference point for pace-setting conversations. A timeline agreed without a shared view of execution capacity defaults to the pace suggested by external narrative. No more specific reference point exists to replace it.
The third concerns accountability language. An eight-point gap in how each party weighs AI performance can be resolved directly, in a single conversation. Left unexamined, it persists as two different operating assumptions inside the same governance structure. The difference becomes visible only once results fall short.
Pace and capacity determine each other in practice. A board and a CEO working from a shared view of execution capacity set a pace the organisation can meet. AI returns then depend on delivery.
Where that shared view is absent, pace continues to outrun capacity. The shortfall becomes visible later, in AI initiatives that do not reach their expected return.