AI is not yet lifting productivity, Barclays tells clients

TL;DR:

• Barclays analysts told clients that evidence AI is making workers more productive remains “unconvincing”, with no clear link between industry-level adoption and faster productivity growth.
• The note lands amid fears of an AI stock market bubble, after US markets fell last week on doubts the infrastructure spend will pay off.
• Just 14% of US workers reported using AI daily for their jobs in the second quarter of 2026.

A Barclays research note has poured cold water on the central promise of the AI boom: that it will make economies more productive. The bank’s analysts found little sign that US industries which embraced AI have seen larger productivity gains than before they adopted it, calling the industry-level evidence “surprisingly fragile”.

A sober counterweight

The timing is pointed. The finding will “fuel concerns about a potential stock market bubble”, the analysts wrote, after US markets crashed last week on fears that the trillions being poured into AI infrastructure will not generate returns. Their read is that adoption is “gradual and steady rather than rapid and transformative”, with most households and businesses still reporting limited exposure.

Real-world caution supports the picture. Uber capped employees at $1,500 (£1,100) in monthly AI coding spend after blowing its budget; Klarna’s chief admitted his AI-driven cost-cutting in customer service “had gone too far” and began rehiring; and Meta conceded mistakes in its AI-led workforce overhaul. Set against that, Federal Reserve chairman Kevin Warsh has called AI “the most productivity-enhancing wave of our lifetimes” — a gap between rhetoric and measured results that Barclays says has yet to close.

The UK reading

For British firms the note reinforces a domestic pattern rather than contradicting it. ONS data this week showed UK businesses using AI widely but shallowly, and early UK labour-market signals are mixed. The productivity dividend that underpins optimistic UK growth forecasts — and the sharp uplift in confidence among UK finance chiefs — is being assumed more than measured.

Looking forward

Barclays does not argue AI will never raise productivity, only that the data does not yet show it. The distinction matters for markets pricing in transformation now. The next test is whether adoption deepens from occasional use into the daily, embedded workflows where measurable gains would finally appear.