Four percent. That is the share of service firms that told the Federal Reserve Bank of New York they had cut jobs because of AI this year. Among manufacturers surveyed, the figure was zero.
Those numbers sit oddly next to how SME owners actually talk about automation. Ask ten of them why a proposal is still sitting unsigned and at least one will eventually admit it is not about the cost or the software. It is that approving it feels like deciding someone loses their job.
A retail distributor lived that exact hesitation for the better part of a year. A quoting-automation tool had been priced, scoped and ready to go, and the owner still hadn't signed off. The person who did manual price lookups all day had worked there for eight years. Approving the tool felt, to the owner, like choosing to let that colleague go, so the file stayed closed while a competitor's faster quote turnaround started pulling deals away one by one.
The hesitation made sense. What it assumed did not hold up.
Goldman Sachs' 2026 survey of 1,256 small businesses found 87% of AI users say the technology augments their staff rather than replacing them, and 93% report a positive overall business impact. A separate 2026 US Chamber of Commerce Foundation survey found just 12% of small businesses call themselves "very likely" to cut staff over AI in the year ahead, and most of that group had not implemented anything yet, meaning the number reflects anxiety more than a plan. The New York Fed's Liberty Street Economics team, tracking real firm behavior rather than sentiment, found businesses roughly three times more likely to retrain a worker because of AI than to lay one off.
What 2026 survey data actually shows
India's own data points the same direction. The Apna MSME Hiring Pulse recorded a 164% year-on-year jump in AI-skill hiring demand inside MSMEs in FY26 — a sector expanding its workforce around automation, not shrinking it.
The distributor eventually signed. The quoting tool went live, and headcount didn't move. The employee who had spent years on manual lookups moved to handling the client calls that actually needed a person on the line, the ones that used to get pushed to the next day because there wasn't time. Quote turnaround dropped from most of a day to under an hour, and the deals that had been slipping to the faster competitor stopped slipping.
The fear versus what happened
None of this means automation never touches a role. It means the SME-scale automation most businesses are actually weighing, invoicing, quoting, scheduling, follow-ups, tends to remove a task from someone's day rather than remove the person from the payroll. The businesses already convinced of that are the ones moving. The ones still waiting for proof are usually looking straight at it.
If a fear like this is what's actually holding back a task you already know should be automated, that's worth a direct conversation before it costs another year in a drawer.