Barclays: AI not lifting productivity yet

On Tuesday, Barclays released a note to clients with a sobering assessment of artificial intelligence's impact on the economy: despite widespread adoption by companies worldwide, AI has not demonstrably boosted productivity in US industries. The bank's analysts stated that evidence adopting the technology makes workers more productive was "unconvincing" and that "industry-level evidence linking AI adoption to stronger productivity remains weak." They characterized adoption as "gradual and steady rather than rapid and transformative," noting that most households and businesses are still reporting limited exposure to the technology.

The finding strikes directly at one of the central justifications for the massive financial commitments companies and investors have made to AI infrastructure. US stock markets crashed last week amid concerns that trillions of dollars being spent on AI would not deliver expected returns. Federal Reserve chair Kevin Warsh has offered a counterweight to such doubts, recently characterizing the AI boom as "the most productivity-enhancing wave of our lifetimes – past, present and future" and arguing that AI would be "structurally disinflationary," lowering the cost of goods and potentially enabling lower interest rates. Barclays' conclusion directly challenges this narrative, stating that such a productivity pickup "remains surprisingly fragile."

Several large companies have already begun pulling back from aggressive AI spending or transformation plans. Uber implemented cost controls earlier this year, capping all employees' monthly token spending per AI coding tool at $1,500 after the company ran through its AI budget. Klarna's CEO admitted last year that the company's pursuit of cost-cutting through AI in customer service had gone too far; the fintech had frozen hiring for more than a year to focus on building AI capabilities before restarting customer service hiring in 2025. At Meta, CEO Mark Zuckerberg acknowledged in recent months that the social media giant had made mistakes during its AI-driven workforce transformation, which included laying off 10% of employees and transferring 7,000 workers to new AI-focused initiatives. These retrenchments suggest that while companies remain committed to AI development, the gap between promised gains and realized results is forcing a more measured approach.

Barclays' finding that AI adoption has not yet boosted productivity in US industries challenges the prevailing optimism that has driven trillions of dollars in corporate and infrastructure spending. The bank's analysts noted that adoption appears "gradual and steady rather than rapid and transformative," with most households and businesses still reporting limited exposure to the technology. This conclusion becomes particularly significant given the high stakes staked on AI productivity gains: Federal Reserve chair Kevin Warsh has recently characterized the AI boom as a path to lower interest rates, arguing it would be "structurally disinflationary" by reducing costs and lowering inflation.

The tension between promised productivity gains and current evidence has already prompted some companies to moderate their AI bets. Uber and Klarna have both had to rein in spending or rehire after pursuing aggressive AI-driven cost-cutting strategies that proved unsustainable or counterproductive. Meta's public acknowledgment that its large-scale workforce transformation tied to AI went awry reflects a similar recalibration. These moves suggest that while enthusiasm for AI's potential remains, companies are discovering that translating that potential into measurable productivity and cost improvements is proving harder than initially expected.

No comments yet. Be the first to share your thoughts!