Atlanta Fed survey shows AI spending among US firms nearly doubled

American companies are spending a lot more on artificial intelligence, and they’re doing it fast.

US firms that use AI raised their spending on it by 96.1% from 2025 to 2026, according to the Federal Reserve Bank of Atlanta’s Survey of Business Uncertainty. That brings the average to $1,024 per employee.

What the Atlanta Fed found

The Survey of Business Uncertainty, run by the Atlanta Fed, polls executives about their plans and expectations. The latest round was conducted from September 14 to 25, 2026.

Executives at companies already using AI said they plan to invest an average of $1,024 per worker for calendar year 2026. That figure represents the 96.1% increase over the prior year.

A broader March 2026 round of the survey painted a similar picture with bigger numbers. In that round, average expected AI investment per employee rose from $1,358 in 2025 to $2,068 in 2026, a gain of approximately 50%.

Scale that across the economy and the estimate for total private sector AI investment comes in close to $280 billion. That is an estimate built from executive expectations, not a receipt.

The average hides a very lopsided picture

The median expected investment for 2026 sits at only $200 to $294 per employee, according to the research findings.

Meanwhile, the top 10% of firms project spending more than $2,800 per employee.

Approximately 70-78% of US firms reported using AI as of late 2025 and early 2026, with larger and more productive companies leading the way.

By sector, professional and business services stands out. Those firms expect to spend about $3,470 per employee in 2026, a 74% rise from the year before.

Executives are spending big but expecting modestly

Companies are pouring money into AI, yet executives aren’t forecasting a productivity revolution.

Respondents predicted modest productivity gains over the next few years. They also flagged potential reductions in employment, especially among workers without a college degree.

That outlook echoes what the survey picked up in 2025. Back then, executives reported limited short-term effects on productivity or employment, balanced by optimism about how efficiently AI could be deployed down the road.

What this means

For investors, the spending trend points to sustained demand for AI products and services. If firms follow through on plans anywhere near the estimate of close to $280 billion, that money flows to software vendors, cloud providers and the startups building on top of them.

When the median firm expects to spend a few hundred dollars per worker and the top tier expects to spend thousands, growth depends heavily on a concentrated group of big buyers.

The gap between spending and expected payoff is worth watching closely. Executives are nearly doubling budgets while predicting only modest productivity gains.

For the labor market, executives anticipate potential employment declines, with non-college-educated workers flagged as most exposed.

The sector breakdown offers a clue about where effects may show up first. Professional and business services, already the heaviest spender at about $3,470 per employee, is the natural place to watch for early evidence of whether AI changes hiring or output.

With adoption around 70-78% but median spending low, the next phase may hinge on whether the typical firm moves from experimenting to investing at scale.