Meta layoffs, AI, and the worsening job market for California tech workers

Source: Los Angeles Times

Published: 2026-05-19

Entity Analyzed: General Knowledge Worker Category

URL SCAN

California tech workers battered by years of mass layoffs were hoping the job market would rebound this year. It is getting worse. AI has triggered fierce competition for top talent while fueling tens of thousands of layoffs. The class divide in Silicon Valley is widening: a tiny group lands unprecedented AI packages while many others struggle. Since 2022, over 815,500 tech workers have been laid off. In 2026’s first four months, U.S. tech employers announced 85,411 cuts — up 33% year-on-year. California information jobs fell 17% from mid-2022 to February 2026. Meta is reassigning 7,000 workers to AI roles while cutting roughly 8,000 others. GM cut 600 IT workers; Walmart is reportedly cutting or relocating ~1,000 tech staff. Hiring cycles have elongated to six months, salaries are dropping, and recruiters demand AI skills for roles that never needed them.

The Triage

This is not a business story. It is a social autopsy with five bodies on the table. The LA Times does what most outlets won’t: it names the dead. Basem Istanbouli, 33, a former Google account manager with years of experience, has been unemployed for over a year despite ‘making it to final rounds very often.’ He started a hiking group called (un)PTO for people in career transition — a community built from the wreckage of an industry that used to guarantee stability. Jason Zhang, 25, a software engineer at Google since 2022 — ‘the dream for most college students’ — got laid off and now documents the brutality on TikTok and Instagram. Kira Martins, 36, lost her digital asset management job at Snap in April. Bruce Bowers, 64, a 40-year veteran at Xerox, Sun Microsystems, and Oracle, got his pink slip and decided retirement was the only rational response.

The numbers behind these names are staggering. 815,500 tech layoffs since 2022. 85,411 cuts in the first four months of 2026 alone — a 33% year-on-year jump. California information jobs down 17%. The San Francisco Bay Area, once the engine of American innovation, now has 0.4% job decline where it used to see 7.5% growth. The triage: the tech labor market is not cycling. It is breaking. The class divide the article names — ‘a tiny group of employees are landing unprecedented packages for their AI skills, while many others struggle to find work’ — is not a temporary distortion. It is the new architecture.

The Autopsy (with DT-LAG)

Mechanical Collapse Point

The mechanical collapse point is visible in the hiring process itself. Robert Lucido at Magnit notes ‘elongated hiring cycles’ — employers are interviewing more people for each role, taking up to six months to fill positions, and the salaries offered to successful candidates are dropping. Paul Flaharty at Robert Half confirms that companies are creating new AI roles while eliminating old ones, and the only path for displaced workers is ‘upskill.’ But upskill into what? The article does not say, because there is no honest answer. A 33-year-old account manager with a decade at Google cannot ‘upskill’ into an AI research scientist in six months. A 64-year-old product manager cannot ‘upskill’ into a machine learning engineer. The ‘upskill’ narrative is the polite fiction that allows the industry to blame workers for their own displacement.

The mechanical reality is that AI is not replacing workers one-for-one. It is redefining what ‘work’ means at the organizational level. Snap’s explanation for laying off Kira Martins is revealing: employees are using AI to ‘reduce repetitive work, increase velocity, and better support our community, partners, and advertisers.’ The framing is that AI makes the remaining workers more productive. The unspoken corollary is that fewer workers are needed. Snap did not say it was automating her job. It said it was using AI to increase velocity. The velocity increase is the layoff.

Lag-Weighted Social Timeline

Immediate (0-6 months): The (un)PTO hiking group is the social form of the immediate lag. Displaced workers are not rioting. They are networking. They are hiking. They are building communities of the surplus. This is the soft landing that policy makers hope for — but it is not a transition. It is a holding pattern. Basem Istanbouli’s ‘final rounds’ that never convert to offers are the lag in action: the interview pipeline is still running, the processes are still formal, but the hiring decisions are not being made. The companies are maintaining the theater of opportunity while the actual opportunity contracts.

Short-term (6-18 months): The 33% year-on-year increase in tech layoffs accelerates. The Meta reassignments — 7,000 workers moved to AI roles while 8,000 are cut — become the template. The message is not ‘we need fewer workers.’ It is ‘we need different workers.’ The workers being cut are not being replaced by AI. They are being replaced by other workers who are willing to work with AI. The short-term social reality is a massive skills arbitrage: workers with AI-adjacent credentials command premium salaries, while workers with legacy tech credentials face a market that treats them as depreciating assets.

Medium-term (1-3 years): The California information job decline of 17% is not a blip. It is a structural shift. The Public Policy Institute of California’s data shows the Bay Area has moved from 7.5% growth to -0.4% decline. That is not a recession. That is a regime change. The medium-term social reality is geographic: the concentration of tech talent in San Francisco, Seattle, Austin — once a moat — becomes a trap. When the industry contracts, the concentrated talent pool becomes a concentrated unemployment pool. Workers who do not relocate face a local market that is oversupplied with their exact skills.

Long-term (3-7 years): The Hemenway Falk/Tsoukalas ‘automation arms race’ model meets the California labor market in its most visible form. Competitive pressure between Meta, Google, Snap, and every other tech firm drives displacement beyond what is individually rational. Each firm cuts workers because competitors cut workers, creating a downward spiral that no single firm can escape without surrendering competitive position. The long-term result is not a smaller tech workforce. It is a transformed tech workforce: heavily weighted toward AI infrastructure, revenue-generating roles, and compliance theater, with the middle tier of operational and support functions hollowed out.

Lag Factors

Interview Theater: The ‘up to six months’ hiring cycle and ‘final rounds’ that do not convert are not friction. They are a lag mechanism. By maintaining the formal structure of opportunity — applications, interviews, final rounds — the industry creates the appearance of a functioning labor market while the actual hiring rate collapses. Basem Istanbouli’s experience is the data point that reveals the theater: he gets interviews, he makes final rounds, he never gets offers. The process exists. The outcome does not.

‘Upskill’ Narrative: Paul Flaharty’s advice — ‘find ways to upskill themselves so that they can make themselves as attractive as possible for these new jobs’ — sounds reasonable. It is not. The ‘new jobs’ require years of specialized training in fields that did not exist five years ago. A 64-year-old Oracle product manager cannot upskill into an AI prompt engineer. A 36-year-old digital asset manager cannot upskill into a machine learning ops specialist in the time frame that unemployment benefits cover. The upskill narrative is a reputational buffer for employers, not a transition pathway for workers.

Geographic Concentration: The Bay Area’s 0.4% job decline is magnified by the region’s status as a tech monoculture. When the industry contracts, there is no alternative employer base to absorb the displaced workers. Bruce Bowers’ decision to retire rather than search for work is rational not because he is old, but because the local market for 64-year-old tech workers is nonexistent. The geographic lag is that workers cannot move as fast as capital can reallocate.

Stock Market Theater: Meta’s stock will likely rise on the 8,000 layoffs. Cisco’s stock surged 20% on its layoffs. The market rewards displacement because displacement signals ‘capital efficiency.’ But the market’s reward is the worker’s punishment. The lag is that stock prices move in milliseconds while displaced workers move in months. The market has already priced in the layoffs before the workers receive their severance letters.

Credential Inertia: Jason Zhang’s ‘Google was always kind of like the dream for most college students’ is the credential lag in human form. Students are still pursuing software engineering degrees at the same rate, still taking on debt, still believing the 2022 hiring pipeline is intact. The pipeline is not intact. It is broken. But the inertia of educational institutions — the multi-year degree cycle, the accreditation process, the curriculum design — means the credential factory keeps producing workers for a market that no longer wants them.

Defensive Moats

Community Networks: The (un)PTO hiking group is a genuine moat. In an oversupplied labor market, personal networks are the only reliable channel to opportunity. The workers who survive are not the ones with the best resumes. They are the ones with the best relationships. The moat is social, not technical.

Financial Cushion: Kira Martins’ statement — ‘I’m lucky in that I worked for a tech company. I managed to put some money in my savings’ — is the moat that separates the displaced from the desperate. Tech workers who saved during the boom years have a runway. Those who did not — the younger workers, the contract workers, the workers with families — face immediate crisis. The financial cushion is a class moat, and it is widening.

Narrative Adaptation: Jason Zhang’s decision to document his layoff journey on social media is a moat of a different kind. By making his displacement visible, he creates a personal brand that may generate alternative income. The workers who survive the transition are not necessarily the ones who ‘upskill’ into AI. They are the ones who pivot out of the tech labor market entirely — into content creation, entrepreneurship, or adjacent industries where their skills have scarcity value.

Physical World Inertia: Real estate, vendor contracts, supply chains, and regulatory requirements still create friction for mass displacement. The friction is why Meta is ‘reassigning’ 7,000 workers rather than firing all 15,000 at once. The inertia does not stop displacement. It merely stretches it across a longer timeline, giving some workers time to adapt — if they use it.

Future-Proofing Scorecard

| Timeline | Score | Commentary |
|———-|——-|————|
| 1 year | 2/10 | Layoffs accelerate. The 33% year-on-year increase continues. Meta’s 8,000 cuts set the template. Hiring freezes spread. The ‘upskill’ narrative collapses under its own impossibility. |
| 2 years | 1/10 | Dual labor market solidifies. AI-adjacent roles at premium. Legacy tech roles in permanent decline. Geographic concentration becomes a trap, not a moat. The Bay Area exodus begins. |
| 5 years | 0/10 | The concept of ‘tech worker’ has bifurcated: AI infrastructure engineers vs. everyone else. The middle tier — product managers, account managers, support engineers — is gone. California information jobs do not recover to 2022 levels. |
| 10 years | 0/10 | The tech industry’s employment model is fully replaced by capital-first allocation. The only roles that survive are those attached to revenue-generating infrastructure or regulatory compliance. ‘Tech worker’ as a stable middle-class category is historical memory. |

The Verdict

This is the most human autopsy of the Discontinuity Thesis yet. The LA Times does not talk about Gartner surveys or stock prices. It talks about Basem Istanbouli hiking with other unemployed tech workers, about Jason Zhang wondering if getting laid off was ‘the best thing that happened’ to him, about Bruce Bowers deciding that 40 years in tech is enough. The verdict is not in the numbers — though the numbers are devastating: 815,500 layoffs, 85,411 cuts in four months, 17% job decline in California. The verdict is in the stories.

The tech industry built its brand on ‘changing the world’ and ‘making information free.’ What it is actually doing is making a specific category of human labor — the educated, credentialed, professional-class worker who believed technology was a ladder — economically surplus. The class divide the article names is the real output of the AI revolution: not universal abundance, but bifurcated scarcity. A tiny group commands unprecedented packages for AI skills. Everyone else joins (un)PTO and hopes the final round converts to an offer.

The verdict: the discontinuity is not technological. It is social. The machines are not replacing workers. The capital is reclassifying workers — from ‘talent’ to ‘cost,’ from ‘asset’ to ‘liability,’ from ‘the dream’ to ‘the redundancy.’ The workers who survive will not be the ones who upskilled fastest. They will be the ones who understood, before the market did, that the game had changed — and chose to stop playing it by the old rules.