Women in AI by FemTechConf

The Gender Gap in AI and Tech: What Wage and Workforce Data Reveal in 2026

A cross-market analysis of the gender pay gap, women's participation in core technology roles and the shrinking female share of the UK AI workforce.

By Sophie Keller, Women in AI Editorial Fellow ยท 5 September 2026

The gender gap in artificial intelligence is often described as a recruitment problem. That is only one part of it.

New analysis from three Meridian Review Group publications shows how the gap appears at several connected stages of working life. In Germany, women earn less on average in the information and communication sector and work fewer paid hours across the economy. In the United States, women are much less likely than men to work in core computer and mathematical occupations. In the United Kingdom, women now account for only one fifth of AI roles even while employers report widespread AI skills shortages.

These are not identical measures and should not be treated as if they are. Together, however, they describe a system in which participation, occupational access, seniority, working time and pay can reinforce one another.

The data at a glance

The three reports identify several important signals:

Germany's unadjusted gender pay gap was 16% in 2025, rising to 19% in information and communication. Germany's broader Gender Gap Arbeitsmarkt, which combines hourly pay, working time and employment participation, stood at 37%. The United States had 37 million STEM workers in 2024, representing 26% of the workforce, but men remained substantially more likely to work in computer and mathematical occupations. Among US workers aged 25 to 34, around 7% of employed men worked in computer occupations compared with around 3% of employed women. Women held 20% of UK AI roles in 2025, four percentage points lower than in 2020. In the same UK survey, 97% of organisations identified at least one AI skills gap and 57% reported a technical skills gap.

The strongest conclusion is not that one number explains the entire gender gap. It is that the gap changes form at different points in the talent and career pipeline.

Germany shows why the wage gap is also a technology-sector issue

German Business Review's analysis begins with the national pay gap but becomes more revealing when it looks at sectors and working patterns.

Women in Germany earned 16% less per hour than men on an unadjusted basis in 2025. The adjusted gap was 6% after accounting for characteristics available in the official statistical model. In information and communication, the unadjusted hourly gap was 19%. Professional, scientific and technical services recorded a 25% gap.

An unadjusted pay gap is not evidence that a woman is paid 19% less than a man for performing the same technology job. It measures the difference between average gross hourly earnings and reflects factors including occupation, seniority, career history and the distribution of women and men across roles.

That distinction does not make the figure less important. It changes the question employers need to ask. If women are less represented in senior technical positions, concentrated in different functions or progressing at a different rate, formally neutral salary bands will not remove the overall earnings gap.

Germany's broader labour-market measure makes the pattern clearer. The Gender Gap Arbeitsmarkt reached 37% in 2025 once hourly pay, paid working time and employment participation were considered together. Women averaged just under 28 paid hours each week compared with a little over 34 hours for men.

For technology companies competing for scarce specialists, this is also a capacity question. A labour market that does not convert women's skills into the same access to paid hours, technical responsibility and senior work is leaving talent underused.

The US data shows where access to high-value technology work narrows

The United States has an enormous science and technology labour market. As American Commerce Review reports, the National Science Foundation estimated that 37 million people worked in STEM occupations in 2024, equal to about 26% of the US workforce.

The scale of the sector does not mean access is evenly distributed.

Bureau of Labor Statistics age profiles show that men aged 25 to 54 were consistently more likely than women of the same ages to work in computer and mathematical occupations. Among employed people aged 25 to 34, about 7% of men worked in computer occupations compared with around 3% of women. Among those aged 45 to 54, the shares were 5.8% for men and 1.9% for women.

This is an occupational participation measure, not a direct pay-gap measure. Its economic significance comes from the value attached to these roles. The same report notes that mean annual wages reached about $148,100 for software developers and $126,800 for data scientists in May 2025.

When women enter these occupations at a lower rate, the effect extends beyond one salary. It can influence retirement savings, equity compensation, routes into management, access to influential professional networks and the experience needed to start or fund a company.

That is why measuring only the gender balance across an entire company can be misleading. An organisation may appear balanced while women remain underrepresented in the technical roles most closely connected to product decisions, AI systems, intellectual property and future leadership.

The UK AI workforce is growing, but women's share is falling

The UK evidence brings the issue directly into artificial intelligence.

According to British Business Review's analysis of the UK Government's AI Labour Market Survey 2025, women held 20% of AI roles in 2025. That was four percentage points lower than in 2020.

There is one positive movement inside the data. The share of surveyed organisations employing no women in AI roles fell from 53% in 2020 to 41% in 2025. Female participation was spread across more employers, but women's total share of AI roles still declined.

A plausible explanation is that the AI labour market expanded faster than female participation. More businesses may have hired at least one woman while the larger volume of new roles remained disproportionately filled by men.

The decline matters because it is happening during a skills shortage. Some 97% of surveyed organisations identified at least one AI labour-market skills gap, 57% reported a technical skills gap and 28% said technical shortages had affected their ability to achieve business goals.

This turns representation into a competitiveness issue. Employers cannot credibly describe AI talent as scarce while treating the underrepresentation of women as a separate problem to address later.

How the gaps compound across a career

The three markets reveal different parts of the same career sequence:

Entry: Who gains the education, confidence, networks and first opportunity needed to enter technology? Occupational access: Who reaches computer, engineering, data and AI roles rather than remaining concentrated in adjacent functions? Project allocation: Who receives technically consequential work that builds promotion evidence? Progression: Who becomes a senior engineer, research leader, manager or executive? Working patterns: Who can remain in high-responsibility roles through caregiving, career breaks or changes in working hours? Economic reward: Who captures the salary, bonus, equity, pension and entrepreneurial upside created by technology growth?

Weakness at one stage changes the pool available at the next. Fewer women entering core technical work means fewer candidates for senior AI positions later. Lower access to commercially important projects can slow promotion. Reduced paid hours or rigid leadership roles can widen lifetime earnings differences even when hourly rates are controlled.

The result is not one gender gap. It is a chain of connected gaps.

Why AI could widen the divide

AI is not simply another part of the technology sector. It is changing the value of work across the economy.

People who build, govern and deploy AI may gain access to fast-growing roles, higher bargaining power and new paths into leadership. People whose work is more exposed to automation may face greater pressure to reskill. If women are underrepresented in the first group and overrepresented in the second, AI adoption could widen existing differences in earnings and career security.

There is also a decision-making consequence. The people who design models, choose data, set product priorities and approve deployment influence which problems technology solves and which risks receive attention. Representation in AI teams therefore affects both access to economic opportunity and the range of experience shaping the technology itself.

What employers should measure now

A single company-wide representation percentage is not enough. Employers need a view of the full system.

Useful measures include:

applicants, interviews, offers and hires by role and seniority; starting pay and subsequent pay progression; representation in software, data, machine learning and AI teams; access to high-impact projects and customer-facing responsibility; promotion velocity across technical and management tracks; retention following parental leave or other career breaks; paid working patterns and the availability of flexible senior roles; representation in succession plans and executive leadership.

The goal is not to produce more reporting for its own sake. It is to locate the point where opportunity narrows and test whether an intervention changes the outcome.

Transparent salary architecture may address one problem. Return-to-work pathways may address another. Sponsorship, visible technical assignments and credible staff-level career tracks can address progression. Education and reskilling partnerships can widen entry routes.

From representation targets to operating decisions

The data from Germany, the United States and the United Kingdom does not support one universal solution. Labour markets, education systems and employment patterns differ.

It does support a more practical principle: women's participation in AI should be managed as part of talent strategy, innovation capacity and economic opportunity, not as a communications category.

Companies already measure product adoption, engineering throughput, hiring time and leadership succession because those outcomes affect performance. Representation across the AI career pipeline deserves the same operational attention.

For individuals, clearer routes into the sector matter as well. Practical learning, portfolio evidence, professional networks, visible role models and access to employers can help turn interest in AI into durable careers. Our AI careers and skills hub provides practical guidance, while our Women in AI research hub tracks representation, careers and leadership.

Building a wider AI economy

The AI economy is still young enough for its career pathways to change. That is an opportunity, but it is not a guarantee.

If AI investment grows much faster than women's participation, existing technology gaps may become embedded inside a larger and more valuable industry. If employers, educators and policymakers treat participation as part of the skills and productivity agenda, the current technology cycle could instead open new routes into technical and leadership work.

The Women in AI Global Summit 2027 is designed to bring those groups together. Researchers, engineers, executives, founders, investors and policymakers need a shared conversation about who builds AI, who leads its adoption and who benefits from the opportunity it creates.

Closing the gender gap in AI requires more than increasing hiring at the entry point. It requires widening access to the entire career, from the first technical role to the highest levels of pay, ownership and decision-making.

Sources and further reading