UK SME AI Spending Lags Behind Infrastructure Hype

Oracle faces delays in its massive data centre expansion due to issues with gas pipeline infrastructure, highlighting a critical bottleneck at the heart of the artificial intelligence boom. While tech leaders and investors project trillions of dollars will be spent on computing power over the coming years, actual adoption rates among businesses tell a different story.

Capital on Tap, one of the United Kingdom’s largest lenders to small and medium-sized enterprises, has released data revealing that AI spending is growing but remains far behind expectations. In the second quarter of 2024, only 3.2 per cent of SMEs purchased services from artificial intelligence companies. By the second quarter of 2026, this figure had quadrupled to reach 12.8 per cent.

The average expenditure on these technologies has risen significantly during this period, moving from £93 for a typical small business in early 2024 to £288 by mid-2026. However, the mean figure is skewed upwards by a very small number of high-spending organisations at the top end of the market. When looking at the median spend instead, which offers a more realistic view for the average company, costs stand much lower at just over £75 per business in the most recent quarter.

Despite this growth, artificial intelligence tools still account for only 0.1 per cent of total card transaction value processed by Capital on Tap. When extrapolated across the wider economy, annual spending on AI services within the UK is estimated to be approximately four billion pounds. This amount represents less than a tenth of Tesco’s annual turnover.

These figures align with broader global estimates suggesting worldwide investment in artificial intelligence ranges between one hundred and one hundred fifty billion pounds annually. Industry analysts note that these levels will likely increase as major providers raise prices for their models. However, this growth occurs against the backdrop of plans by the world’s largest technology firms to invest nearly a trillion dollars on infrastructure over the next two years.

Much of this proposed investment is funded through debt rather than immediate revenue generation. For such massive infrastructure bets to become profitable, business spending on artificial intelligence would need to increase exponentially beyond current trajectories. While adoption rates are accelerating rapidly, they have not yet reached the scale required to justify the sheer volume of capital being deployed into data centres and computing hardware.

This discrepancy raises significant questions for investors who must reconcile the optimistic projections from technology giants with the modest reality faced by everyday businesses. The gap between projected infrastructure costs and actual market uptake suggests that the economic model underpinning the current AI boom faces substantial challenges before it can sustain itself without further massive debt accumulation or a dramatic shift in corporate spending habits.

The situation underscores the difficulty of making complex financial sums add up when technological hype outpaces practical application. Investors are increasingly pondering how to navigate this landscape where billions are committed to building capacity that may not yet have sufficient demand to support it.