AI’s economic boom is already showing some awkward cracks

AI is expanding quickly, but the clearest evidence so far may be the contrast between a vast infrastructure boom and less settled signs of everyday adoption and business demand. Derek Thompson’s chart-led review sets out both sides, including estimates of $280 billion in annualised external AI revenue and signs that spending is concentrated among a small group of customers.

Watch Desk analysis

What happened

Thompson’s 26-chart review of AI and the economy brings together figures on public use, business spending, infrastructure and revenue. It says roughly half of Americans report using AI chatbots, while daily use for work is much lower, at 10% to 25%, depending on the survey.

The review also says that AI infrastructure spending is highly concentrated: Apollo data cited by Thompson puts more than 90% of model and data-centre spending with the top 10% of AI customers. Separately, Ramp’s chief economist reported that the top 1% of businesses on its platform cut AI spending per employee by 10% in August. Thompson notes that this could reflect seasonal factors, a wider shift, or both.

Why it matters

The figures describe different parts of the same question: is AI becoming useful enough to support the enormous sums going into its infrastructure? Rapid growth in revenue and investment can coexist with uneven workplace use and spending that depends heavily on a small group of customers. That makes headline growth a poor substitute for asking who is paying, how often they use the tools and whether the benefits last.

The review also estimates external AI revenue at about $280 billion annualised as of September, drawing on Exponential Growth. That is a striking figure, but it is an estimate, not a settled industry-wide tally. Thompson points to falling token prices and possible shifts towards cheaper models as factors that could complicate the revenue picture.

Our read

This is a useful map of an economy-sized argument, not a verdict that AI is either a bubble or a miracle. The reader payoff is the tension: investment and reported revenue are immense, while usage and spending are uneven. Keep those measures separate, and be wary of anyone using one chart to declare the whole case won.

What to watch

• Whether business AI spending broadens beyond the biggest customers.
• Whether daily workplace use grows alongside the infrastructure build-out.
• How falling model prices affect revenue and the economics of AI products.

Discussion spark: If AI spending stays concentrated among a small group of businesses, is that evidence of a young market finding its footing, or a warning that the boom rests on too few customers?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.