AI use is spreading rapidly, but much less is known about businesses spending on AI. This column uses new survey evidence to estimate total AI spending in 2026 at around $280 billion in the US and $25 billion in the UK. Spending per employee is expected to increase by more than 40% in both countries compared with 2025, driven by younger, more productive, and higher-paying firms. Yet expenditure remains highly concentrated: fewer than 2% of firms account for half of total expected AI spending in each country.
Artificial intelligence (AI) tools are spreading rapidly in much of the economy. Large AI technology firms are investing heavily in data centres and the infrastructure needed to train large language models (Carpinelli et al. 2026, Van Nieuwerburgh 2026, Soto et al. 2026). Adoption is widespread among businesses more generally: survey evidence from over 9,000 executives across seven countries show that 78% of firms now use at least one AI technology (Yotzov et al. 2026).
As AI use becomes more common, understanding how intensively firms use these technologies becomes increasingly important. Spending provides one measure of intensity. In this column, we present new survey evidence on AI-related expenditures by American and British firms in 2025 and their expected spending in 2026. The results show rapid spending growth, alongside striking differences across firms and a high degree of concentration.
Our evidence comes from two large, economy-wide business surveys: the US Survey of Business Uncertainty (SBU), run by the Federal Reserve Bank of Atlanta, and the UK Decision Maker Panel (DMP), run by the Bank of England. Both surveys collect information from senior business executives. Alongside questions on the AI technologies their firms use and the realised and expected effects on employment and productivity (Yotzov et al. 2026), we asked respondents about AI-related spending.
The questions covered expenditure in calendar year 2025 and expected expenditure in 2026. For 2026, we asked: “Approximately how much do you expect your firm will spend on artificial intelligence adoption and AI related technologies? Include software, subscription services, hardware, worker training and IT staff support.” The question about 2025 asked how much the firm spent in the previous year. Our broad spending concept captures purchases of AI technologies and the spending needed to support their adoption within firms.
We fielded the SBU questions in March 2026, yielding 845 responses, and the DMP in August–September 2026, yielding 1,273 responses. We divide reported spending by firm-level employment to quantify AI expenditure per employee and winsorise spending at the 99th percentile to limit the influence of outliers. We convert UK responses from pounds sterling into US dollars at a rate of $1.35 per pound.
Figure 1 shows the distribution of AI spending per employee in 2025 and expected spending in 2026 for US firms (Panel A) and UK firms (Panel B). In both countries, firms expect AI expenditure to become more widespread and spending per employee to rise.
More than 80% of US firms reported some AI-related spending in 2025, and 23% spent more than $1,000 per employee. Only 16% of firms report no AI-related spending in 2025, and 11% in 2026. The employment-weighted average across all US firms rises from around $1,363 per employee in 2025 to an expected $2,043 in 2026, an increase of roughly 50%.
These averages conceal a highly skewed distribution. Median spending in the US is just $294 per employee in 2026, while the top 10% of firms expect to spend an average of $7,480 per employee. Extrapolating from the survey responses gives an estimate of approximately $280 billion in AI spending across the US economy in 2026.
Panel B shows a similar pattern of expanding expenditure among UK firms, although spending per employee is lower than in the US. One-third of UK firms report no AI-related spending in 2025, but this share is expected to fall to 17% in 2026. The share spending between $100 and $999 per employee is expected to rise from 29% to 44%, while the share spending more than $1,000 is expected to increase from 8% to 13%.
Average AI spending per employee in the UK is expected to rise from $422 in 2025 to $725 in 2026, an increase of approximately 72%. Extrapolating these responses to the UK economy gives an estimate of around $25 billion in total AI-related expenditure in 2026.
Figure 1 AI spending per employee
AI spending varies systematically with firm characteristics. Figure 2 presents four binned scatterplots relating expected AI spending per employee in 2026 to firm-level measures of productivity, pay, and age. We use UK data for this analysis, where these firm-level measures are readily available.
More productive firms expect to spend significantly more on AI per employee (Panel A), as do firms that pay higher average wages (Panel B). Expected spending is also higher among firms with younger directors (Panel C) and younger firms (Panel D).
Figure 2 AI spending per employee is greater at younger, more productive, higher-paying firms
There are also large differences across industries, as Figure 3 illustrates. US firms in the finance, information and professional services sectors expect to spend over $3,000 per employee in 2026, compared to under $1,000 per employee for manufacturing firms (Panel A). In the UK (Panel B), AI spending per employee in 2026 is also expected to be highest among firms in the finance, information, and professional services sectors (around $1,300); other sectors expect to spend less than $1,000 per employee.
Figure 3 AI spending per employee by industry
This figure is based on responses to the question: For calendar year 2026, approximately how much do you expect your firm will spend on artificial intelligence adoption and AI related technologies? Include software, subscription services, hardware, worker training and IT staff support. Panel A is based on 845 responses from firms in the US Survey of Business Uncertainty in March 2026 and Panel B is based on 1,273 responses to the Decision Maker Panel in August-September 2026. The results are weighted by firm employment. Original responses in Panel B are in GBP and converted to USD using an exchange rate of 1.35 $/£.
Finally, Figure 4 shows how concentrated AI spending is across firms. It plots the cumulative share of total expected AI spending in 2026 against the cumulative share of firms. In both countries, a very small share of firms accounts for a large share of expenditure: just 1.8% of US firms and 1.3% of UK firms account for half of all expected AI spending in their respective countries. This concentration is consistent with evidence from AI suppliers. For example, Anthropic’s initial public offering (IPO) documents showed that one-quarter of its sales were to just two firms (Financial Times 2026).
Figure 4 Firm AI spending is highly skewed
How unique is this concentration of AI-related spending? In Figure 5, we plot the distributions of sales, employment, and investment across UK firms using accounts data and across US firms using Compustat for the 2024 financial year. Based on these figures, AI spending is more concentrated than employment or sales, but about equally as concentrated as total firm investment. This comparison suggests that the result is not simply an artefact of the survey data: AI expenditure appears to share the highly uneven distribution characteristic of firm investment more generally.
Figure 5 Distribution of sales, employment, and investment
Our survey evidence shows that US and UK firms expect rapid growth in AI-related spending from 2025 to 2026. In both countries, average spending per employee is expected to increase by more than 40%. Extrapolating from our survey responses implies economy-wide spending of about $280 billion in the US and $25 billion in the UK in 2026. The rapid rise in AI spending suggests that AI usage intensity is also on the rise, as many firms invest heavily on AI technologies and their integration into business processes.
Spending varies considerably across businesses. In the UK data, younger, more productive and higher-paying firms expect to spend more per employee. Strikingly, expenditure is highly concentrated: fewer than 2% of firms account for half of total expected AI spending in 2026 in the US and the UK. As adoption continues to spread, it will be important to track both the growth of AI spending and whether this concentration declines as more firms begin to use the AI technologies more intensively.
Source : VOXeu
Should newly arrived refugee children immediately enter mainstream schools or first attend separated schools? Using…
Europe’s innovation gap with the US and China lies in breakthrough rather than incremental innovation.…
High public debt around the globe has revived concerns that it could undo central banks’…
Politicians have long pressed central banks for lower interest rates in the belief that easy…
From employer wage-setting power to the consequences of pay transparency and the role of benefits…
Bankers may benefit from the upside of risk-taking through bonuses, while taxpayers may ultimately bear…