California employers with 100 or more workers submitted their Reporting Year 2025 pay data to the Civil Rights Department by May 13, 2026 — the most comprehensive mandatory gender pay disclosure the state has required of its tech sector. The aggregate picture is not a surprise, but the legal framing around it has shifted materially. Women in computer and mathematical occupations earn approximately 83 cents for every dollar men earn, a 17-cent gap that Payscale's 2026 Gender Pay Gap Report occupational breakdown translates to $14,300 in annual median earnings for that category. What is new in August 2026 is the mechanism that makes that number visible: Google, Meta, and Microsoft stopped publishing voluntary DEI reports in 2025. NVIDIA removed its DEI section from its 2026 10-K. California's SB 1162 annual pay data filing is now, for most frontier AI employers, the only mandatory public record on what women in their companies actually earn relative to men.
What the CRD Aggregate Data Shows
On March 26, 2026, the California Civil Rights Department released its annual analysis of Reporting Year 2024 pay data — drawn from reports covering nearly eight million workers across large California private-sector employers. The aggregate findings carry a pattern that has been consistent since SB 1162 took effect: women are concentrated in lower-paid job categories, and the managerial inversion compounds with seniority. Across the filing universe, men hold 63 percent of senior manager and executive positions; women hold 37 percent. In the category classifications covering most tech and AI work — professional and technical roles, computer and mathematical occupations — the CRD's aggregate data documents a persistent gap between women's median hourly rates and men's.
The CRD does not publish individual employer filings. What it releases is sector-level aggregate data by job category and sex — the dataset that is publicly queryable through its results portal. The Reporting Year 2025 cycle, filed in May 2026, continued to use the original 10 EEO-1 occupational categories; AI safety, machine learning research, and model evaluation remain collapsed into the "Computer and Mathematical" bucket in the aggregate data the CRD published in June 2026. SB 464, signed into law in 2024, expands the framework to 23 Standard Occupational Classification categories beginning with Reporting Year 2026 data, due May 2027. That expansion will produce the first job-category-specific gender pay gap figures for AI-specific roles — not tech broadly, but the narrower slice that frontier labs fill.
The DEI Pullback Raises the Stakes
The voluntary DEI reporting retrenchment by major tech companies in 2025 is the context that gives the CRD filings their new weight. Google, Meta, and Microsoft confirmed they would not publish their annual diversity, equity, and inclusion reports — a reversal of more than a decade of public reporting. NVIDIA deleted its DEI section entirely from its 2026 annual 10-K filing; a shareholder proposal filed in June 2025 specifically cited the reduced workforce data transparency as a governance concern. The result is that employers who previously offered voluntary public disclosure on gender representation and pay equity have quietly exited that disclosure regime — while their SB 1162 obligations remain unchanged.
The company that stands apart is Salesforce. Since 2015, Salesforce has conducted annual global pay equity audits and published the results. In its most recent assessment covering 2024, 3 percent of employees received upward pay adjustments after the company's analysis identified unexplained disparities. Salesforce has spent approximately $3 million per year in audit-driven remediation since the program began. That process has produced a specific, checkable track record of where gaps were found and what was spent to close them — a standard none of the frontier AI labs currently meet.
Anthropic has not published a public diversity or pay equity report. The company's research organization grew from approximately 280 to 720 between Q1 2024 and Q1 2026, placing it squarely inside the SB 1162 annual reporting requirement. The RY 2025 filing, submitted this May, is the first year in which Anthropic's pay data — aggregated at the sector level in the CRD's published findings — contributes to the public record at meaningful scale.
The AI Skills Premium Multiplies the Gap
The structural gender pay gap in tech is not primarily a within-role pricing problem. Payscale's 2026 report, drawing on surveys of more than 130,000 US workers, places the controlled gender pay gap — adjusted for job title, level, experience, and location — at $0.99 per dollar. The problem is role assignment, not within-role pay. Women are systematically underrepresented in the roles that carry the highest pay premiums, and the fastest-growing premium in tech is for AI skills specifically.
PwC's 2025 Global AI Jobs Barometer documented a 56 percent wage premium for AI skills. McKinsey and LeanIn's Women in the Workplace 2025 report found that only 21 percent of entry-level women are encouraged by their managers to use AI tools, compared with 33 percent of men at the same level — a 12-percentage-point gap in access to the skill-building that feeds the premium. When employees are actively encouraged to use AI, they are more than 50 percent more likely to do so. The compounding effect is direct: women are less likely to be steered toward AI tools, which reduces AI skill development, which reduces access to AI-premium roles, which widens the total compensation gap that shows up in SB 1162 filings.
Women currently hold approximately 22 percent of AI roles globally and 18 percent of machine learning engineer positions. Among AI researchers specifically, the figure falls to 12 percent. At the frontier lab layer — where ML engineers and research scientists earn $280,000 to $560,000 in total compensation — the representation gap and the compensation premium compound simultaneously. The gap between what the posted salary range shows and what the CRD annual data reveals is partly the mathematical output of those two forces running in the same direction.
Company Profiles: A Gap in the Record
Google's technical workforce is 19.1 percent women against a 34.2 percent overall representation — a gap that reflects exactly the role-assignment dynamic Payscale's controlled data isolates. Google stopped publishing its voluntary DEI annual report in 2025. Its SB 1162 filing for RY 2025 remains the binding public record on what women in its California technical workforce earn relative to men. Meta filed a 2025 gender pay report covering its Ireland operations — a jurisdiction that requires it. That filing showed a mean hourly pay gap of negative 3.2 percent (women earning slightly more per hour in the mean), but a median bonus pay gap of 89.9 percent. The bonus gap is where AI seniority competes: equity grants, performance bonuses, and profit-sharing structures that scale with role level and AI adjacency. Base hourly parity coexisting with a 90 percent bonus gap is not pay equity. It is the specific accounting signature of a company where women are represented at lower levels of the AI org and men dominate the bonus-eligible senior tier.
NVIDIA's 2022 ESG data placed women at 19 percent of its total workforce; the company has since claimed annual pay parity through internal review while reducing the visibility of those claims by removing DEI disclosures from its regulatory filings. What remains is the SB 1162 filing and, separately, the state's aggregate data — neither of which NVIDIA controls.
Using This Data Before Signing an Offer
Three data sources now exist that AI candidates can combine in compensation negotiations, and the combination is more powerful than any single source.
The first layer is the CRD's public-facing results portal. The aggregate median hourly rates by job category and sex are queryable by industry and occupation. A candidate in a computer and mathematical role can confirm where the sector-wide gender median sits for their job category before walking into an offer conversation. If the CRD's aggregate data shows women in that category earning a median hourly rate materially below the posted salary floor in the relevant job posting, that divergence is a documented data point — not a negotiating instinct.
The second layer is Levels.fyi self-reported total compensation, which shows gender breakdowns in some role categories and provides a distribution view of where actual offers land relative to posted ranges.
The third layer is the posted range floor itself, which under SB 1162 and Colorado's EPEWA is a legally committed minimum. Colorado's CDLE had received more than 2,800 complaints, issued over 600 voluntary compliance letters, and collected more than $800,000 in penalties as of June 2026. A candidate who receives an offer below a company's posted California floor has a straightforward enforcement referral pathway.
The CRD data, the Levels.fyi distribution, and the posted floor are three different instruments measuring the same underlying compensation structure. Using all three simultaneously — rather than treating the posted range as the only number that matters — is the practical adaptation that the SB 1162 annual reporting regime makes possible.
The filings are in. The aggregate data is public. Whether a candidate uses it as a floor for negotiation or a footnote to skim is the only remaining variable.
Methodology and sources: California Civil Rights Department, "New Data Shows Ongoing Gender and Race Pay Divide Among Large Private Sector Employers," March 26, 2026 — calcivilrights.ca.gov. Payscale, 2026 Gender Pay Gap Report — payscale.com/press-releases/2026-gpgr. McKinsey and LeanIn, Women in the Workplace 2025, December 2025 — leanin.org/report/women-in-the-workplace. AAUW, The Simple Truth About the Gender Pay Gap, 2025 edition — aauw.org/resources/research/simple-truth. Colorado Department of Labor and Employment EPEWA enforcement data through June 2026 — sixfifty.com Colorado pay transparency overview. Meta Ireland Gender Pay 2025 Report — s21.q4cdn.com Meta Ireland filing. Salesforce 2024 Annual Equality Update — salesforce.com/news/stories/annual-equality-update-2024. PwC Global AI Jobs Barometer 2025. NVIDIA DEI disclosure reduction — governance-intelligence.com. DEI reporting pullback — allwork.space. Women in AI representation data: edX Women in AI report 2025 — edx.org/resources/closing-ai-gender-gap. California SB 1162 filing deadline and 2026 requirements — stoelrivesworldofemployment.com. All company compensation data: Levels.fyi through August 2026; Glassdoor salary submissions through July 2026.
Global Career PlatformFind AI talent. Find your next role.
Booking is hotels. · Airbnb is apartments. · ENTRA is global careers.