Two claims about artificial intelligence and the size of a company have been circulating this month, and on the surface they contradict each other. Agenda Digitale argues that generative AI is tilting the field toward small, nimble organizations. A wave of fresh research says the opposite is happening at large companies: firms that cut staff in the name of AI are not seeing the productivity payoff they expected. Both can’t be describing the same phenomenon, or can they?

What the lean-startup argument actually claims

Stefano da Empoli, president of the Institute for Competitiveness, makes the case in Agenda Digitale that AI rewards flexibility more than scale. Generative tools have pushed down the cost of building competitive technology, so a compact team no longer needs a large budget or a deep bench of specialists to ship something genuinely good. A small organization also carries less internal friction: fewer approval layers, shorter decision chains, room to pivot the moment a tool or a market signal changes. The argument stays light on hard numbers and named examples, but the underlying logic holds up: AI removes a cost barrier that used to protect incumbents, and it rewards whoever can act on a new capability fastest.

The layoffs that didn’t pay off

A separate body of research, summarized by The Conversation and covered by Dataconomy, tells a less flattering story about big companies. Researchers combed through millions of Glassdoor reviews, thousands of corporate filings, and hundreds of AI-investment and layoff announcements from U.S. public firms over five years. The pattern: companies that announced large AI investments also tended to announce AI-linked job cuts, yet those cuts rarely delivered the promised productivity gains. Stock markets barely reacted to the layoff announcements, which suggests investors weren’t buying the efficiency claim either. The scale is not trivial: more than 122,000 tech workers lost their jobs in 2025 and over 126,000 in 2026, and close to 29 percent of companies that trimmed staff ended up reopening the same roles later at salaries 20 to 35 percent higher, with total rehiring costs running one and a half to two times what they thought they’d saved.

Why fear undermines the efficiency story

The most telling detail in that research has nothing to do with the technology itself. Employees who stayed after a round of AI-linked layoffs grew wary of the tools that had just cost their colleagues their jobs, and that anti-AI sentiment tracked with lower firm-wide productivity. An Atlanta Federal Reserve survey cited in the same body of work found that roughly 90 percent of executives believe AI has not yet improved productivity at their own company. Sentiment, in other words, mattered more than any optimism coming from the top. Cutting people to fund an AI rollout tends to poison the very adoption it was supposed to accelerate, a dynamic I touched on when writing about tech layoffs and redundancies.

What I’ve actually seen: freed time, not fewer people

My own experience with AI points to why these two accounts aren’t as contradictory as they look. Claude handles the repetitive, rule-bound slice of my work: pulling sources, drafting a first pass, checking a fact, formatting a table. What comes back to me is time, and I spend that time on the part no tool can do for me: deciding which argument actually holds, what to leave out, which tone fits a given article. That’s a redistribution of effort toward judgment and creativity, not a smaller headcount. The Agenda Digitale piece and the layoff research are, in a sense, describing two different choices companies make with the exact same technology. A lean team has nowhere to put the hours AI frees up except back into the product, the writing, the strategy, because there’s no spare organizational fat to absorb them. A large company can choose instead to bank that saved time as a cost cut, and that choice is precisely what the productivity studies are measuring the damage of.

So who’s right?

Both camps, partially. Agenda Digitale is correct that AI lowers the cost of competing, and small organizations are structurally better positioned to convert that saving into speed, because they tend to reinvest freed capacity rather than eliminate it. The layoff research is equally correct that treating AI as a justification for cutting people, rather than as a way to redeploy them toward higher-value work, produces fear, resentment, and worse output, exactly as I explored when looking at AI and the labor market. Size isn’t really the deciding factor here. What decides the outcome is whether the hours AI saves go toward more thinking or toward fewer people doing the thinking. Small companies tend to choose the former almost by necessity. Large ones too often choose the latter, and the numbers above suggest they’re paying for it.

Related sources

• Piccolo è bello nell’era dell’AI: la rivincita delle startup snelle (Agenda Digitale)
• Layoffs tied to AI hurt worker productivity, and the reason may surprise managers (The Conversation)
• Study Finds AI-driven Layoffs Are Failing To Deliver Productivity Gains (Dataconomy)
• Layoffs tied to AI hurt worker productivity, and the reason may surprise managers (TechXplore)