Employees at 90% of Firms Use AI — But Only 40% of Firms Pay for It
Only 40% of surveyed firms pay for LLM subscriptions, yet staff at 90% use AI anyway. A CEO of an 800-person company says start with goals, not tools.
By Olivia Hart
3 min read

What's News
- Employees at 90%+ of surveyed companies report regular AI use; only 40% of firms have official LLM subscriptions.
- Cited research claims AI can absorb 60-70% of employee busywork.
- AI tool market estimates range from ~2,500 to more than 50,000 products.
- BBVA empowered process experts, not central IT, to design AI solutions, per HBR.
- Author leads a company of more than 800 employees with weekly AI demo days.
Employees at more than 90% of surveyed companies report regular use of artificial intelligence, yet only 40% of those companies have purchased official subscriptions to large language models, according to a recent report cited by software executive leadership. That gap — workers adopting AI faster than their employers sanction or support it — sits at the center of a new management argument: stop chasing tools and start with business goals.
The case comes from a CEO leading a company of more than 800 employees, writing in Fast Company under the headline "Why the smartest AI strategy starts with a little laziness." The author's definition of laziness is selective: not idleness, but the hunt for a faster route. The piece leans on a line attributed to Bill Gates: "I choose a lazy person to do a hard job. Because a lazy person will find an easy way to do it."
How much busywork can AI actually absorb?
The author cites research suggesting AI can absorb 60% to 70% of busywork, freeing employees for strategizing and problem-solving. That is a large claim, and the source offers no timeline or methodology. Even taken at face value, the figure implies the bigger constraint is managerial, not technological — piecemeal, unsanctioned adoption means companies miss the compounding benefit.
The remedy proposed is top-down clarity. The author points to estimates that the AI tool market contains anywhere from roughly 2,500 to more than 50,000 products, noting the lower figure is likely closer to reality for live, available tools. Either way, no full-time employee can test them all. The takeaway in plain words: pick the problem first, and only then the tool.
One cited example is IBM's Watson, which the author says has been used in healthcare to analyze patient data and medical literature and give doctors evidence-based treatment options.
What does employee-led adoption look like?
The author's company runs weekly demo days where any of the 800-plus employees can present AI tools they have discovered — customer service reps, for instance, advising designers on how to better program help chatbots. The logic echoes a bakery analogy: if you were building cooking technology, you would ask the baker first, not the product designer.
The piece also quotes a Harvard Business Review article on the European bank BBVA: "Instead of relying on the core IT and operations functions, the organization empowered the people who understood business processes best to design the solutions." BBVA's experience, as reported, suggests frontline knowledge outperforms central IT mandates for process-level AI — though the article does not quantify BBVA's results, a caveat worth holding onto.
The practical prescription is a standing channel — an online forum, a Slack thread, or a weekly all-hands — where employees contribute views on AI development and share best practices.
Why treat AI spend like a portfolio?
Every AI initiative competes for time, budget, talent, and executive approval, the author writes, recommending that employees build slack into their schedules to test tools and recapture minutes of busywork. A second HBR article cited in the piece recommends managing AI investments like a portfolio, evaluated against objective criteria: alignment with business goals, training needs, required resources, and the cost of diverting them. "The goal is to ensure that individual projects advance business strategy and that the overall mix of investments creates sustainable value," the HBR article states.
The author's closing claim is that success with AI depends less on which tools companies adopt and more on their willingness to stay focused — to be, in the piece's phrasing, "lazy about how much you take on, and smart about what you choose." That is a slogan, not evidence. But the underlying data point stands: with AI use running at 90%-plus of firms against 40% official licensing, the unpaid question for executives is how long they can let shadow adoption set their AI strategy for them.
By Olivia Hart
Original: hbr.org
More from Olivia Hart
Show full bio
Staff writer covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.
2,300 articles