Bank of England pours cold water on AI productivity boom

The Bank of England estimates AI has lifted UK productivity by 0.5% over the past three years, adding just 0.1–0.2 percentage points to annual growth since early 2024; AI-related activity accounted for just under a fifth of UK GDP growth since the beginning of 2026, largely through development and deployment rather than productivity gains from adoption. Survey evidence also points to limited realized returns: Gartner found 3% of organizations had achieved their AI productivity and cost targets, while 38% were over budget and 85% planned to spend more next year. The BoE expects AI to add around 0.9 percentage points to annual productivity growth over the next three years, but cautioned that AI-related demand for products such as memory chips is raising import costs for Britain.

Analysis

The investable signal is a split between AI infrastructure demand and customer return on investment—not evidence that AI demand is about to collapse. For the next 1–3 months, weak realized productivity could pressure valuation multiples for firms priced on rapid enterprise monetization, while persistent spending keeps near-term demand intact for compute and memory suppliers. Those effects can coexist; rising component prices may support upstream revenue but squeeze adopters’ budgets and delay deployment economics.

The UK evidence should not be extrapolated mechanically to US or global productivity: it is a narrow read-through, and adoption gains may lag installation. The more consequential risk is that customers continue spending without measurable returns, then redirect or cap budgets when renewal decisions arrive. That is a 6–18 month catalyst, not yet proof of an imminent capex break. Conversely, broad-based productivity gains or improving customer ROI would weaken the bear case.

For Gartner (IT), the survey evidence is not automatically negative: weak AI execution can sustain demand for research and decision support, but it does not establish revenue conversion. Watch subscription growth, renewals and management commentary before treating the survey as an earnings signal. The inflation channel is also asymmetric: higher AI-related import costs could complicate UK disinflation, but the article does not quantify pass-through enough to justify a rates position. Overall, this is a caution against paying peak multiples for downstream AI monetization, not a standalone short signal on the AI complex.

Key Decisions for Investors

• Avoid adding to high-multiple AI application/software exposure solely on promised productivity gains; require evidence in customer renewals, realized cost savings or guidance before increasing risk.
• Do not short AI infrastructure solely on this UK productivity reading. Treat memory and compute pricing as a separate near-term demand signal; reassess if hyperscaler spending plans or supplier order commentary weaken.
• Keep Gartner (IT) on watch rather than making a directional trade from the survey. Verify subscription and renewal trends and whether AI-related research demand translates into reported growth.
• Monitor UK import-price and inflation data over the next 1–3 months for evidence of AI hardware costs feeding through. A sustained rise in inflation expectations would strengthen the case for caution on UK duration; absent pass-through, no rates trade.