California's AI tax windfall masks job losses and bubble risk, analysts warn

California’s fiscal dependence on AI-driven stock gains leaves the state exposed on two fronts: the possibility that the boom is a bubble, and the documented reality that it has not produced the employment growth that past tech cycles delivered.

California is collecting a growing share of its income-tax revenue from tech companies riding the artificial intelligence boom, but a new state analysis warns the windfall is concentrated among a small number of workers and could evaporate if the AI market stalls. Tech companies’ stock-option withholding made up roughly 10% of all California income-tax withholding in 2025, up from more than 6% just three years ago, according to an analysis by the Legislative Analyst’s Office. The gains have come without broad job growth — and as the state braces for a nearly $18 billion budget deficit.

The AI sector’s financial gains have not translated into employment growth. Tech jobs in the Bay Area declined over the past year, separate analyses show, and economists and market analysts are sharply divided over whether the boom is sustainable.

A narrow revenue base

Chas Alamo, the Legislative Analyst’s Office’s principal fiscal and policy analyst, based his estimate on public financial filings from the state’s five most valuable tech companies by market value — Apple, Google, Nvidia, Broadcom and Meta — along with Intel, Cisco, AMD, Intuit, PayPal, Applied Materials and Qualcomm.

Several of those companies posted sharp stock gains in 2025. Google shares rose 59% for the year, Broadcom gained 46%, and Nvidia climbed 25%.

“We’re seeing a real boost to income-tax receipts because of this — for a relatively small number of employees,” Alamo told CalMatters. “If the AI market were to deteriorate, we could see these withholdings decline.”

California’s largest revenue source is personal income tax. Tech companies commonly pay employees in stock options in addition to base wages; vested options are taxed as ordinary income, generating withholding payments to state and federal governments. The concentration of that income among a small number of workers makes the revenue stream unusually sensitive to swings in tech-sector valuations.

The timing of the LAO analysis is pointed. California is expected to run a nearly $18 billion deficit this year, and the state anticipates having to fill additional funding gaps caused by cuts from the Trump administration. The Legislative Analyst’s Office has been raising concerns about “the stagnant nature of the state’s labor market and broader economy” for the past two years, Alamo said.

Jobs picture

The stock-market gains have not produced hiring. Bay Area tech employment declined even as AI investment surged. Jobs in the information industry fell 1.3% from September 2024 to August 2025, and professional and business services fell 1.5% over the same period, according to an analysis by the Bay Area Council Economic Institute, a think tank supported by the Bay Area Council business coalition.

The California Center for Jobs and the Economy, the research arm of the California Business Roundtable, documented a loss of more than 130,000 high-tech jobs — including manufacturing positions — through the first quarter of last year. Salesforce, based in San Francisco, cited AI as a factor when disclosing layoffs affecting thousands of employees.

“Right now, on net, AI is not a job-gainer,” said Jeff Bellisario, executive director of the Bay Area Council Economic Institute. “Tech booms in the past have led to an employment boom. This doesn’t feel like that.”

California’s unemployment rate stood at 5.6% in September 2025, the most recent data available, making it the highest among U.S. states, according to the LAO.

Bubble debate

Analysts disagree on whether the investment boom will prove durable. Jensen Huang, chief executive of Nvidia, has pushed back on skeptics. “There has been a lot of talk about an AI bubble,” Huang told investors in November. “From our vantage point, we see something very different.”

Dan Ives, managing director at Wedbush Securities, said the comparison to the dot-com era misreads the current moment. “This is not a bubble,” he said. “This is Year 3 of an eight- to 10-year buildout of the AI revolution.”

Others are less certain. Analysts at Allianz Trade, the global insurance company, wrote in a November brief that “the financial market frenzy over AI shows classic signs of an asset bubble: widespread consensus, unproven valuations and returns at times detached from earnings.” PitchBook, which tracks public and private capital markets, warned in its 2026 outlook that some AI subsectors — including AI scribes for health care, aerial defense drones, content development in gaming, and personal assistant bots — contained too many startups, and that firms would need to differentiate themselves to sustain value.

Policy stakes

California leads all states in AI regulation efforts and is expected to resist a Trump executive order that would establish federal AI rules superseding state laws.

Kaitlyn Harger, an economist for Chamber of Progress, a think tank funded by the tech industry, argued the state has financial reasons to be cautious about regulation. “What’s important to remember is that California’s social safety net depends on a healthy tech industry,” she said.

The analysis by the LAO did not recommend a policy course. It documented the dependence and named the risks — a sector that has enriched a small class of workers and generated substantial tax revenue, with little evidence that the gains are spreading, and meaningful uncertainty about how long they will last.