NFIB May Survey: AI Capex Booms While Small Business Hiring Hits 6-Year Low

If you wanted a single chart to capture the strangeness of the AI economy in mid-2026, the National Federation of Independent Business just handed you one. The NFIB's Small Business Optimism Index dropped 0.6 points in May to 95.3 — its latest leg down, and still stuck below the survey's 52-year average of 98.0. Meanwhile, the same report's accompanying commentary acknowledges what everyone reading this publication already knows: AI investment is the thing keeping the broader economic mood from curdling entirely.

"AI investment spending has contributed to some excitement in the economy," said NFIB chief economist Bill Dunkelberg in the release. The rest of his statement was considerably less cheerful, pointing to a divided picture in which small firms are getting squeezed by volatile fuel costs they can't pass along to customers the way their corporate competitors can.

Strip out the framing and the data is blunt. Hiring plans among small business owners just hit their lowest level in six years. A seasonally adjusted net 9% of owners say they intend to add jobs over the next three months, down 4 points from April and the weakest reading since May 2020 — which, you'll recall, was the month the entire economy was locked in its house. The share of owners reporting unfilled job openings fell 5 points to 29%, also a six-year low. That's not a labor market cooling gracefully. That's small businesses quietly stepping back from the hiring table while hyperscalers pour hundreds of billions into datacenters a few tax brackets away.

The Two-Speed Economy Gets a Number

For the past two years, AI watchers have debated whether the capex supercycle — the GPU clusters, the power deals, the construction cranes — was masking softness everywhere else. The May NFIB report, first flagged in coverage by Investing.com, is about as clean a data point for that thesis as you'll get. The survey's Uncertainty Index climbed 3 points to 91, towering over its historical average of 68. Small business owners aren't pessimistic so much as paralyzed: they don't know what input costs, demand, or policy will look like two quarters out, so they're sitting on their hands.

And sitting on their hands is exactly what the capital spending numbers show. Just 16% of owners plan capital outlays in the next six months, down a point from April and the lowest reading since March 2009 — the bottom of the global financial crisis. Let that sink in: while Big Tech's AI infrastructure spending is the single largest driver of fixed investment growth in the national accounts, the median small firm is investing at depths last seen when Lehman Brothers was a fresh wound. The aggregate investment numbers that make GDP look healthy are being carried almost entirely by a handful of balance sheets that can afford to bet on compute.

Prices, meanwhile, are going the wrong way. A net 36% of owners raised average selling prices in May, up 6 points and the strongest reading in some time, while 70% reported supply chain disruptions touching their business — a 6-point jump from April. Falling hiring intentions plus rising prices plus record-low investment plans is a combination economists generally file under 'unpleasant.'

The Labor Cost Signal AI Companies Should Be Staring At

Buried in the report is the number that matters most for anyone building or selling AI products: 14% of small business owners now name labor costs as their single most important problem — up 5 points in one month and the highest reading in the survey's entire history. At the same time, complaints about labor quality fell to 13%, the lowest since December 2016.

Read those two lines together and a story emerges. Small businesses are no longer struggling to find competent workers; they're struggling to afford them. That is, almost word for word, the pitch deck of every AI agent startup founded since 2024. When the cost of human labor is your customers' loudest pain point and the availability of it isn't, the substitution math for automation gets easier to sell. Vertical AI companies targeting bookkeeping, customer support, scheduling, and back-office work for sub-50-employee firms have spent two years arguing that SMBs are the real market for agents. The NFIB just published their best top-of-funnel slide for free.

The skeptical read is just as available, though. A small business that won't commit to capital outlays at 2009-crisis levels is not obviously about to sign an annual SaaS contract for an AI receptionist. Frozen spending is frozen spending. The record labor-cost complaint creates the motive for AI adoption; the record-low investment intent removes the means. Which force wins probably depends on pricing — usage-based AI tools that show up as a variable cost rather than a capex line have a structural advantage with this exact buyer right now, and vendors who haven't figured that out are selling into a wall.

Why the AI Boom Isn't Trickling Down — Yet

There's a genuinely open question about transmission here. The hundreds of billions flowing into AI infrastructure do create jobs and regional spending — electricians, concrete, grid upgrades, the much-discussed datacenter construction boom. But that activity is geographically lumpy and sectorally narrow. The NFIB's membership skews toward the restaurants, contractors, retailers, and service firms that feel fuel prices and supply chains long before they feel any spillover from a hyperscaler's Virginia campus.

Dunkelberg's fuel-price point cuts deeper than it first appears. "More small business owners are struggling with significant and unpredictable hikes in fuel prices," he noted, and unlike large corporations, they can't easily push those costs downstream. There's an AI angle even here: the datacenter buildout's strain on energy markets has been a simmering policy story all year, and if electricity and fuel volatility is partly a side effect of compute demand, then the AI boom isn't just failing to lift small businesses — it may be actively taxing them through the power bill. That causal chain is contested and hard to isolate, but expect it to feature in the political fight over datacenter siting and energy pricing heading into the fall.

The employment index itself tells the slow-bleed version of the story: 100.3 in May versus 100.4 in April, the third straight monthly decline, now sitting below its 2025 average of 101.2. Nothing is collapsing. Everything is sagging.

What to Watch From Here

Three threads worth tracking. First, whether the labor-cost record translates into measurable SMB AI adoption — watch earnings and usage disclosures from Intuit, Shopify, and the agent startups that report anything at all, because they're the canaries for whether 'AI for Main Street' is a market or a meme. Second, the June and July NFIB prints: one bad hiring number is noise, but hiring plans and openings hitting six-year lows simultaneously starts to look like a trend, and a fourth straight employment-index decline would harden the case. Third, the macro debate itself — if small business hiring keeps rolling over while AI capex keeps the headline numbers respectable, the 'is AI masking a recession' argument moves from podcast fodder to Fed testimony.

The AI industry has spent a year insisting its spending boom benefits everyone. The 33 million small businesses in America just filed their response, and it reads less like enthusiasm than an invoice.