Enterprise AI’s Bottleneck Is No Longer Technology
BCG’s AI at Work survey of nearly 12,000 workers points past models and tools: the real drag is strategy, organization, and people.
Ask how an enterprise should “do AI,” and the first answers are usually familiar: we need a bigger model, more tools, more tech spend. BCG’s latest AI at Work survey of nearly 12,000 workers across fourteen markets suggests the opposite. Adoption has already moved far. What lags is strategy, organization, and how people are managed.
The “silicon ceiling” cracked
For years, AI looked like a leadership and tech-team privilege: frontline white-collar staff barely touched it, or used it poorly. The industry nickname was the “silicon ceiling.” This year’s numbers say that story is outdated. About 74% of frontline white-collar workers are now regular AI users—daily or several times a week—up roughly 23 percentage points from 2025. In one year, AI moved from a managerial habit toward a general workplace tool.
Regionally, India, the Middle East, and Australia lead frontline adoption—India near 95%, the Middle East and Australia around 93%. Markets often assumed to be “tech ahead,” such as the United States, France, and Italy, sit behind the global average in that cut. By function, IT (about 88%) and marketing (about 85%) lead; finance, analytics, and risk roles also run above average. Sales and operations lag, with operations near 61%.
When everyone uses AI, the hard question changes
That shift is symbolic. The old friction was willingness and skill. The new friction is what happens next: where does saved time go, how do job designs change, and does management catch up? Roughly 72% of respondents say skill expectations for their role have already shifted because of AI. About 67% say AI has taken most simple, repetitive work, leaving harder and riskier tasks. Around six in ten say the bar for “good enough” work is higher—because AI raises the baseline.
Nearly half say their center of gravity has moved from doing the work by hand to directing AI—prompting, editing, deciding. Review time rises; decision work rises. AI is becoming a collaborator, while many org charts still treat it like a personal plug-in.
The second half is organizational
Read together, the survey’s signal is blunt. The first half of enterprise AI—getting tools into people’s hands—is largely done. The second half is not who has the strongest model, but whose strategy is clear, whose workflows were redesigned, and who can combine technology with people. For individuals, “can I open the tool” will soon be table stakes; the edge is managing AI, making harder decisions with it, and still showing judgment that is hard to copy.