Conventional wisdom holds that the youngest entrepreneurs are the most comfortable with new technology. A May 2026 analysis of millions of small-business banking accounts complicates that assumption, and it reframes the women-owned business AI adoption gap as something more surprising than a simple age curve. The JPMorgan Chase Institute studied de-identified transaction data from Chase Business Banking deposit accounts spanning 2019 to 2025, tracking which firms actually paid for artificial-intelligence tools. The finding is not just that men are ahead, but that the gap is widest among the youngest owners — a pattern that points to confidence and trust rather than capability.
The numbers behind the women-owned business AI adoption gap
By the end of 2025, roughly 19.7 percent of male-owned businesses had adopted AI tools, compared with 17.2 percent of women-owned businesses. That difference looks modest until it is tracked over time: the gap widened from just 0.3 percentage points in 2019 to 2.5 points in 2025 — more than eight times larger. Because the study follows actual payments to AI services rather than survey responses, it captures spending behaviour rather than stated intent.
The relative picture is more encouraging than the absolute one. The male-to-female adoption ratio peaked at about 1.28 in 2023 and fell to roughly 1.14 by the end of 2025, meaning women-owned firms are catching up proportionally even as the raw percentage-point gap grows. The gap also holds across sectors. It is widest in food services and accommodation, where men adopt at well over 1.5 times the rate of women, and narrowest in transportation and other services — a consistency that, the authors argue, points to structural barriers rather than the mix of industries women tend to run.
The Gen Z paradox
The headline twist is generational. Millennial owners lead overall, reaching 22.1 percent adoption by 2025, with Generation Z close behind at 18.6 percent and well ahead of Generation X and Baby Boomers. Yet the gender gap is not smallest among the digital natives — it is largest. Breaking 2025 adoption down by both gender and generation, the Institute found a 6.1-point gap among Gen Z owners (20.0 percent of men versus 13.9 percent of women), compared with 3.5 points among Millennials and just 0.5 points among Baby Boomers. In other words, “digital native” does not automatically translate into “AI adopter.” Growing up with smartphones is not the same as feeling licensed to bet a young company on an unfamiliar, fast-changing tool.
Why the gap persists
The researchers point to structural reasons rather than any difference in ability. Women-owned firms face tighter access to capital — female founders still receive an estimated 1 to 2 percent of total US venture funding — which limits the budget for experimentation. Women also report greater concern than men about AI data privacy, security and trustworthiness, and those concerns are reasonable rather than a deficit: when the margin for a costly mistake is thinner, caution is rational. Survey evidence reinforces the pattern. A Harvard Business School meta-analysis covering more than 140,000 people across 18 studies found women had about 22 percent lower odds of using generative AI than men, and separate polling shows a wide trust gap by age, with far more older owners than younger ones saying they simply do not trust AI. This mirrors a broader pattern covered in the persistent gap in women holding AI’s top roles: the issue is rarely interest or talent, and almost always access, resourcing and trust.
What this means for women-owned businesses
Closing the women-owned business AI adoption gap does not require a moonshot. It rewards a disciplined, low-risk approach to bringing AI into real workflows. Starting with one bounded problem — a single recurring task such as drafting client follow-ups or reconciling invoices — and measuring the hours saved before expanding keeps the risk small and the evidence concrete. Trust can be treated as a feature rather than a barrier: choosing tools with clear data-handling policies, turning off model training on business data where possible, and keeping a human approval step turns the same caution that slows adoption into an advantage in client-sensitive work.
Cost need not be a wall either. Many capable tools are free or inexpensive to pilot, and a no-code agent builder can test an idea without hiring a developer. As subscriptions multiply, though, so do bills, so watching rising AI subscription costs and cancelling what goes unused matters as much as adopting in the first place. The Institute itself emphasises that access alone is not enough; targeted, use-case-specific training and peer mentorship networks are what tend to build the confidence the data shows is missing.
Limitations and what to watch
The transaction-based method is a strength, but it has edges worth noting. It infers owner gender and age from account signers and could classify firms with only a clear majority, so a slice of businesses falls outside the analysis. Paying for an AI service is also not the same as using it well — adoption measured at the point of purchase says little about depth of integration, which is increasingly what separates firms that see a return from those that do not. The most important signal is forward-looking: because the gender gap is widest in the youngest cohort, and Gen Z is a growing share of new business formation, the disparity could widen as that generation’s firms mature unless confidence and trust are addressed directly.
From gap to advantage
The data tells a counterintuitive story: the generation most fluent with technology shows the largest gender divide in putting AI to work. That makes the gap less a question of skill than of confidence, capital and trust — all of which are addressable. For women-owned businesses, a deliberate, well-governed approach to adoption is not just a way to catch up; the same caution that has slowed uptake can become a selling point in work where clients care how their data is handled. Source: JPMorgan Chase Institute; Harvard Business School.
