The Reality Check

That sums up the conundrum business leaders have faced since AI exploded into the popular consciousness four years ago. AI budgets are rising, pilots are multiplying, and tools are in employees’ hands. But provable gains have often been delayed, diffuse, or dependent on changes outside the technology.

Across all five dimensions of this report — return on investment, workforce and culture, agentic commerce, governance and risk, and sovereignty — we found the same pattern time and again: AI capability is arriving much faster than the decisions required to turn it into value.

To succeed in the next phase of AI, organizations will need to make the difficult choices that many have deferred. Among them, leaders must:

Decide what return each AI investment is meant to produce, and who will capture it.

We find that the ROI question is usually being framed too broadly, applied too narrowly, and asked too late. There is no single, magical ROI number to be discovered or defended; rather, targeted, workforce, foundational, and transformational investments all follow different paths to return and should be judged accordingly. Every major AI investment needs an ROI harvester: a senior leader accountable for turning the capacity AI creates into returns.

Decide what AI will do before asking the workforce to transform around it.

We see a “Great Leadership Misalignment” that needs immediate correction. Leaders must agree on where AI is taking the enterprise and what role employees will play in that journey. If leadership doesn’t write the AI workforce story, employees might assume they aren’t part of it. Only 37% of executives and 11% of non-managers said their company’s leadership has stated a clear AI ambition and is keeping them informed about progress. And many workers worry they will become the savings, with 68% saying they would give up part of their salary for a guarantee against AI replacement.

Decide how the business will win when machines help customers choose.

AI isn’t only changing companies — it’s also changing the behavior of clients and consumers. Already, 48% of Generation Z consumers and 40% of millennials turn to AI first when looking for a new item without a specific brand in mind. When recommendations conflict, respondents are more likely to follow AI than any other source — ahead of friends and family, ratings and reviews, social media, or familiar brands. To reach these consumers, brands must scramble to master the developing science of what we call shopping agent optimization, or SAO.

Decide how to govern AI at the speed at which it operates.

Stronger AI risk management and governance can accelerate growth rather than inhibit it. Where approved AI is easy to reach, clearly governed, and built into everyday software, 85% of workplace AI users report productivity gains, versus 44% where those conditions are largely absent. Sturdy guardrails reduce ambiguity by telling employees which use cases they can pursue, which ones require additional review, and who owns the decision. Better brakes allow racecars to move faster.

Decide which AI dependencies are acceptable, and which are worth paying to control.

AI adoption has become AI dependency. No country or enterprise controls the full stack; the basic inputs — computing power, energy, models, data, identity, tools, and agent operating environments — create layers of interdependence that organizations need to manage. So far, only 35% of workers say their company has put the ability to switch AI providers in place. The practical question for companies, and nations, is where greater control over these factors justifies additional cost, and what that control looks like.

All of these reality checks boil down to the same management challenge: How to harness a technology that is creating options faster than organizations can decide what to do with them. The technology will keep improving whether or not leaders make those decisions correctly. The returns will not.