Labor is a record cost. Nobody's investing to fix it.

Labor has never been a bigger cost problem for American small business. And the share of owners planning to invest in anything at all sits at a level last seen in March 2009 — the depths of the subprime mortgage crisis.

Those two facts cannot both be optimal. Expensive labor is the most reliable historical trigger for capital investment: when people cost more, businesses buy the equipment, install the system, automate the repetitive work. Main Street is doing the opposite.

01Labor stopped being scarce and started being expensive

The number almost everyone skipped explains the rest. Owners citing worker quality as their top problem fell five points to 13%, the lowest since December 2016. Owners citing labor cost rose five points to a record 14%. In one month, they traded places.

Fig. 01 — The crossover

Share of owners naming each as their single most important problem.

Chart by Pulse Business Finance · Data: NFIB SBET, May 2026

Unfilled job openings also dropped five points, to 29% — the lowest since May 2020. Translate that out of survey language: you can find people now. You just can't afford them.

That loosening cuts both ways, and the second edge is easy to miss. The senior operator or controller you couldn't get at any sensible price for three years is available right now. Cycles where good people are available and nobody is bidding against you tend to be short.

02So why is nobody investing?

Fig. 02 — The capital strike

Of every 100 owners, how many plan a capital outlay in the next six months.

Chart by Pulse Business Finance · Data: NFIB SBET, May 2026

The retreat is not irrational. Borrowing costs are still elevated — as we covered when the Fed held again, the posture is higher-for-longer, so capital isn't cheap either.

The bigger reason is the one behind The Cost of Standing Still: the NFIB Uncertainty Index sits at 91 against a long-run average of 68. Owners aren't saying the investment is bad. They're saying they can't underwrite it.

You cannot change the rate environment. You can absolutely change whether you have a model that tells you what an investment survives.

03Waiting is a position, and it has a price

Deferring feels conservative. Usually it just moves the cost somewhere that never appears on an invoice. If labor cost is your top problem and you defer the investment that reduces labor intensity, you haven't avoided the cost — you've committed to paying it monthly, at whatever rate the market sets, indefinitely.

A wage line grows indefinitely. A financed asset amortizes and then stops. "We'll wait until things settle down" is a financial position with a real cost, and almost nobody prices it before choosing it.

That example is the traditional shape of the decision — a large asset, financed, paid back over years. It is also exactly the shape that uncertainty blocks. Which is why the next part matters.

04Where defined AI changes the shape of the decision

NFIB named this tension directly: Chief Economist Bill Dunkelberg noted that AI investment spending has generated real excitement, but the overall picture is divided. That division is the story. The AI capital boom is happening — just not on Main Street.

Here is why the gap is more closable than in previous cycles. Substituting capital for labor used to mean a large, irreversible, hard-to-underwrite asset — precisely the decision uncertainty paralyzes. Defined AI and automation carry a different risk profile entirely.

So the specific thing blocking investment largely dissolves. But the word doing the work in that heading is defined — and the evidence on what happens without it is unambiguous.

Fig. 03 — The adoption gap

Small businesses are using AI. Very few have made it structural.

Chart by Pulse Business Finance · Data: Goldman Sachs 10,000 Small Businesses, early 2026

Most owners are using the tools. Most feel good about them. Very few can point to where the benefit landed. That gap is where the money quietly disappears — not because the tools don't work, but because feeling faster is not a financial result.

05What "well defined" actually looks like

The difference between an automation that pays and one that just adds a line item is almost never the tool. It's whether anyone named the outcome before buying. The loop we keep on the whiteboard:

Fig. 04 — The measured momentum loop

If you can't fill in step four before you spend, it isn't an investment.

Framework: Pulse Business Finance · Measured Momentum series

Run that loop and the economics separate fast. The same tool, at the same price, produces three completely different outcomes depending only on whether the capacity gets converted.

Fig. 05 — Three years, one tool, three outcomes

Cumulative P&L impact of a $500/month tool replacing 10 hrs/week at $38/hr loaded.

Illustrative model by Pulse Business Finance · Not client data

Same subscription. Same vendor. A roughly $59,000 three-year swing between the business that named the outcome and the business that bought the tool and hoped. The tool didn't decide that. Management did.

06What to do this quarter

1. Re-diagnose the labor lineIf you're still spending against a talent shortage when you have a cost problem, you're funding the wrong answer.
2. List your three most labor-intensive processesHours, fully loaded cost, and whether the work genuinely requires judgment. That's your investment shortlist — and your AI pilot shortlist.
3. Name the outcome before the toolOne number, one owner, one deadline. If you can't state what should move and by how much, you're not ready to buy.
4. Price the deferralPut the annual cost of not acting directly beside the cost of acting. Make the comparison explicit.
5. Convert the capacity deliberatelyRecovered hours become money only through a decision: take on more work, avoid a hire, cut overtime, redirect to selling, or collect faster.
6. Arrange credit before you need itIf rates hold, access matters more than price. That's the work we do on the Debt & Growth Capital side.

A record share of owners say labor cost is their biggest problem. The smallest share since 2009 plan to do anything structural about it. That gap isn't irrationality — it's a forecasting gap, and it closes with a model rather than better conditions.