AI productivity is about to become visible and investable

Unlock the Editor’s Digest for free

The writer is head of European equity strategy and head of global derivatives strategy at UBS

Much of the debate around artificial intelligence and productivity has so far been theoretical, or focused on inputs rather than outcomes. Investors have understandably concentrated on capital expenditure, infrastructure build-out and the companies enabling the AI supply chain. What has been harder to observe, and easier to dismiss, is whether AI is yet changing the way businesses operate. That may be starting to change.

In macroeconomic data, productivity has not accelerated. This has fuelled some commentators to say that AI is another overhyped technology cycle whose benefits will take years to materialise. But this framing risks missing a more idiosyncratic shift.

Aggregated data smooths away exactly the effects that matter most for markets: dispersion between relative winners and losers. Investment returns from AI productivity gains are unlikely to arrive evenly across economies or sectors given different use cases, AI investment rates as well as starting profit margins and valuations.

We think it is likely that the most important early signal we can see in company data will be improvements in sales per employee and operating margins. Improvements will be especially important in sectors where margins are thinner and where modest efficiency gains can have an outsized impact on profits. Here are some examples of where AI is already moving the needle for some companies:

Retail & Consumer: Walmart’s AI-driven supply chain automation has enabled up to a 30 per cent reduction in unit costs at fulfilment centres, allowing the company to grow its top line without adding headcount. In many retailers, AI is making employees more productive rather than redundant, reallocating labour to higher-value, customer-facing activities.

Some content could not load. Check your internet connection or browser settings.

Financials: Banks are among the most data-rich institutions and are already seeing incremental, persistent productivity gains. JPMorgan has identified 450 AI use cases in areas such as customer personalisation, trading, and fraud management. Bank of America’s digital assistant “Erica” has handled billions of client interactions, reducing call centre volumes by 40 per cent. These improvements are not yet dramatic, but they are persistent. For banks with large fixed-cost bases, even small improvements in cost-to-income ratios can drive disproportionately large increases in earnings.

Industrials: Deere’s “See and Spray” technology uses AI to reduce usage of some chemicals by 60 per cent, and the company expects 10 per cent of equipment sales to be from recurring services by 2030. Industrial equipment supplier Grainger uses AI to optimise inventory and customer service, and reports a 2.5 percentage point improvement in service levels. In aerospace, Rolls-Royce has saved £180mn on sourced products through AI-driven decision models.

Technology & Services: In software and IT services, AI is already delivering measurable cost savings and margin expansion. SAP’s AI tools have delivered €300mn in efficiencies in 2025, expected to rise to €500mn. In payments, companies such as PayPal and FIS have seen 10 to 30 per cent increases in developer productivity from GenAI coding assistance.

The biggest stock market winners may come from unexpected places such as businesses operating with low margins and heavy labour intensity. A modest uplift in sales per employee or a small reduction in unit costs can translate into a disproportionately large rise in earnings.

Recommended

This is why the next phase of AI-driven returns could be less about the often expensive and crowded beneficiaries of AI spending and more about the often cheaper and less well-owned companies. Their advantage will show up quietly in margins, cost-to-income ratios, revenue per employee and, over time, in sustained positive earnings surprises and valuation expansion.

If investors wanted to look for the data-rich, service-oriented, cost-heavy businesses investing heavily in AI and seeing tangible gains, we think European banks fit the bill nicely. Despite remarkable performance in recent years as interest rates rose from below zero per cent to boost their margins, valuations remain low and yet strong earnings growth continues to be revised higher as costs are consistently trimmed.

AI productivity is unlikely to arrive with fanfare. But markets have a habit of noticing when numbers start to move. For some companies we think they already are and that this effect will broaden market gains throughout 2026.