Most companies aren't using AI. They're trying it.
The median American business spends $10.66 per employee per month on AI. New analysis reveals a commitment gap — not a technology gap — behind AI's missing productivity gains.
- AI
- strategy
- marketing
- business
The median American business spends $10.66 per employee per month on AI. The Economist published that figure this week, and it's worth sitting with longer than the big capex numbers usually get.
The big numbers: America's largest tech companies will spend $900bn on AI infrastructure this year, rising to $1.4trn in 2027. To generate returns on that capital through identifiable income requires revenue of roughly $2.5trn annually — more than tech's entire combined revenue today. Current total AI services revenue is somewhere between $150bn and $220bn annualised. That gap is the main event.
But $10.66 explains why the gap exists.
How thinly AI is actually used
About 20% of American firms say they use AI in any business function, according to the Census Bureau. After peaking at 46% of the workforce in mid-2025, around 33% of workers now use AI on the job. The European Central Bank found that 1 in 10 euro-area companies using AI does so "intensively." German firms: roughly half use AI for 5% of working hours or less. The average American executive uses it for 1.7 hours a week — enough time to produce a decent slide deck, not enough to change how a business runs.
Nine in ten executives report no impact of AI on their firm's productivity over the past three years.
That last figure is the one worth dwelling on.
A commitment gap, not a technology gap
The Economist frames the problem in economics terms: to actually change a business, AI requires about $5 to $10 of complementary investment in process, data, and organisational redesign for every dollar of compute. Call it the intangible work.
That work isn't happening. American companies' investment in organisational capital — processes, routines, data flows, supplier relationships — has been declining as a share of GDP. Data-organisation firms have revenues in the billions, not hundreds of billions. Token usage is growing faster than revenue, which means more people are cycling through free models, not committing to paid ones. Close to half of British businesses using AI don't pay for it at all.
So the picture is: surging infrastructure spend, flattening adoption, minimal commitment, negligible productivity gains, and declining investment in exactly the process redesign that would make AI useful. This isn't a technology gap. It's a commitment gap.
What commitment actually looks like
What commitment looks like is harder to package than a pricing tier. It means reworking how teams operate, how data is organised, how performance is measured — not once, but continuously. Historically, for every $1 spent on computer hardware, companies made $5 to $10 of these intangible investments alongside it. That pattern hasn't shown up yet with AI.
The companies likely to break through aren't treating AI as a $10-a-month subscription. They're treating it as infrastructure that requires substantial complementary work to yield returns, and they're making that work a priority.
The firms spending $10 per employee aren't getting bad technology. They're getting exactly what they're paying for.
Sources: AI revenues are growing fast, but not fast enough — The Economist, July 28 2026. Data cited from the US Census Bureau, European Central Bank, Bundesbank, Bank of England, Ramp, and the Bank for International Settlements.
If you want that question answered for your specific situation, the Forge Playbook does it. Answer a few questions about your business and we'll put together a tailored outline of which workflows are worth automating and what a realistic budget looks like for each. Free, no obligation, takes about three minutes.
