AI Companies Are Now Laying Off AI Workers. Here's What We're Seeing.

The headline: AI is eating itself

Two weeks ago, we wrote that AI had become the go-to layoff justification. Block, HP, Wise, all cutting thousands and citing "AI-driven efficiency."

This week, the story shifted. The AI companies are cutting their own people.

xAI laid off 500 employees from its coding division on March 14. An AI company. Cutting coders. The reason: Grok is falling behind Claude Code and Codex in the AI coding race, and Elon Musk brought in auditors from SpaceX and Tesla to clean house. Two more co-founders were pushed out, leaving only 2 of the original 12. The company that raised billions to build AI is now firing the people who build AI.

Vercel replaced 9 out of 10 salespeople with an AI agent. Three engineers shadowed their best sales rep for six weeks, built an agent to replicate the workflow, and achieved the same lead-to-opportunity conversion rate. The nine displaced reps were moved to outbound sales. The build required one engineer working 25-30% of their time.

Pinterest is cutting up to 15% of its workforce to "reallocate resources to AI-focused roles." That's roughly 700 people out of 4,666. The restructuring will cost $35-45 million. CEO Bill Ready says the platform has become "an AI-powered shopping assistant for 600 million consumers." The humans who built it are being shown the door.

Instacart cut 27% of its workforce. More than 250 positions, primarily in engineering and product. The same teams that built the product are being restructured out of it.

Two weeks ago, companies were using AI as an excuse to cut costs. Now the AI companies themselves are cutting the people who build AI. The snake is eating its tail.

Consumer brands enter the blast radius

The first signals article focused on tech and SaaS. This week, the layoff wave hit retail, DTC, food, and entertainment.

• 🔴 Mattel: 15% of its workforce. The company behind Barbie and Hot Wheels is restructuring across the board.
• 🔴 Glossier: one-third of its staff. 50-54 of 170 employees, gone on February 12 under new CEO Colin Walsh.
• 🔴 H&M: 181 employees in Chicago. WARN notice filed February 20, effective May 7.
• 🔴 Papa Johns: 7% of corporate workforce plus 300 store closures. 200 stores in 2026, 100 more in 2027. Expected to save $25 million.
• 🔴 Delivery Hero (Glovo): 750 workers in Spain. Scaling back services in 60+ locations to avoid closure.
• 🔴 Envato: up to 200 jobs. Nearly one-third of the Shutterstock-owned company, explicitly citing "AI-driven restructuring."
• 🔴 Jellysmack: layoffs at Court TV after acquiring it from Scripps. Dozens of staff including on-air anchors.

"Stable consumer brand" used to mean something in a job search. If Mattel, H&M, and Glossier are all cutting in the same two-week window, that assumption needs updating.

The fintech bloodbath

The financial services and crypto sectors had a brutal two weeks.

• 🔴 Chime: nearly 40% workforce reduction. Headcount dropped from over 10,000 to under 6,000 in late February.
• 🔴 Gemini: 200 jobs (25% of workforce) on February 5 as part of restructuring and market exits.
• 🔴 OKX: global restructuring with undisclosed cuts across the organization.
• 🟠 ZipRecruiter: CFO departed after 10 years, revenue declining, negative profitability.
• 🟠 Gossamer Bio: stock plunged 79.75% after a failed Phase 3 study.

Chime is the standout. Cutting nearly half your workforce in a single quarter is not optimization. It's an emergency.

Where the money is actually going

Not all the signals were red. Defense-tech and AI infrastructure pulled even more capital this week than two weeks ago. The pattern is accelerating.

• 💰 Anduril: $20 billion Army contract. Not a round. A contract. Defense-tech is no longer a niche.
• 💰 Saronic: $600M for autonomous naval vessels. Led by Elad Gil.
• 💰 Apptronik: $520M+ for humanoid robotics, bringing total Series A to over $935 million.
• 💰 Lambda: $480M Series D at nearly $2.5 billion. GPU cloud for AI training.
• 💰 ElevenLabs: $500M led by Sequoia, valuation now $11 billion. Voice AI.
• 💰 Decagon: $250M Series B at $4.5 billion. AI customer support, tripling its valuation.
• 💰 Harvey: reportedly raising $200M at $11 billion. Legal AI.
• 💰 Gecko Robotics: $125M Series D at $1.8 billion. Industrial inspection robots.
• 💰 Fluidstack: $100M from Google at $7.5 billion. Distributed GPU cloud.
• 💰 ClickHouse: valued at $15 billion amid AI analytics demand.

Billions in wages cut. Billions in capital deployed. The money isn't disappearing. It's moving from headcount to infrastructure. From people to platforms. The companies raising money are building the tools that the companies cutting people are buying.

Red flags worth watching

Some signals don't fit neatly into a category but matter if you're evaluating these companies.

• 🟠 CoreWeave: stock dropped 18%. Q4 loss of $0.89/share versus an expected $0.49. Q1 2026 guidance of $2 billion came in $290 million below expectations. They're planning $30-35 billion in capex this year. High ceiling, high risk.
• 🟠 BuzzFeed: going concern warning. Net loss of $57 million in 2025. "Substantial doubt about its ability to continue as a business." If you're interviewing there, read the filing first.
• 🟠 Allbirds: closing nearly all US full-price stores. Revenue down to $33 million in Q3 2025. Market cap: $32 million. The brand is worth less than many Series A startups.
• 🟠 Highspot: merged with Seismic under Seismic's leadership after two layoff rounds and a valuation collapse from $3.5 billion to an undisclosed (much lower) number.
• 🟠 Norwegian Cruise Line: activist investor Elliott took a 10% stake. Stock sinking amid volatility. Activist pressure usually means restructuring is coming.

Other signals from this week

The full scan caught 112 signals across 1,568 companies. Here are the rest worth noting.

• 🔴 TripleLift: one-fifth of its workforce. Over 100 employees in the ad-tech firm.
• 🔴 Genius Sports: 25% workforce reduction while exiting international markets.
• 🔴 Chan Zuckerberg Initiative: 70 jobs (8%) to refocus on AI-powered biomedical research.
• 🔴 Flexport: 2% workforce reduction as part of a push toward profitability.
• 🔴 NielsenIQ: restructuring program targeting $55-65 million in annualized cost savings.
• 🔴 Wolt: up to 45 international operations roles in Finland.
• 🔴 Lucid Motors: 319 workers (12% of US workforce) to improve efficiency amid ongoing losses.
• 🔴 Reitmans: 75 employees at headquarters and distribution centre in Montréal.
• 🔴 FanDuel Sports Network: 118+ jobs across Atlanta, Connecticut, and St. Louis as regional offices shut down.
• 🔴 CyberArk: 500+ employees cut by Palo Alto Networks one day after closing the $25 billion acquisition.
• 🔴 Mercari US: mass layoff months after a major fee structure shakeup.
• 💰 Vir Biotechnology: collaboration with Astellas worth up to $1.7 billion. $335 million in upfront and near-term payments.
• 💰 Janux Therapeutics: collaboration with Bristol Myers Squibb worth up to $850 million.
• 💰 Overland AI: $100 million led by 8VC for autonomous ground vehicles.
• 💰 General Catalyst: raising approximately $10 billion across AI, defense, climate, and healthcare.

What this means for your job search

These aren't just headlines. They change how you should evaluate every company on your list right now.

Rethink what "stable" means. Two weeks ago, we flagged Walmart, Salesforce, and ServiceNow cutting thousands while profitable. Now Mattel, H&M, Glossier, and Papa Johns are doing the same. Brand recognition and revenue are not job security. Before you interview anywhere, look at whether the company is investing in your function or consolidating it.

The Vercel story is a preview of what's coming. One engineer, working part-time for six weeks, replaced a 10-person sales team. If your role involves qualifying leads, triaging support tickets, or processing data, you need to ask in every interview: "How is this team using AI today, and what's the plan for the next 12 months?" If the interviewer doesn't have a clear answer, that's either a company that hasn't started (you'll build it) or one that hasn't decided (you might be the one replaced). The difference matters.

Follow the capital, not the brand. Anduril, Saronic, Apptronik, Lambda, Harvey, Decagon. These aren't household names. But they have fresh capital and aggressive hiring plans. The same pattern held two weeks ago with Cerebras, Waabi, and ElevenLabs. The safest bet right now isn't the biggest logo. It's the company with runway, momentum, and a reason to grow your team.

Know which of your tasks are exposed. Anthropic's research showed 75% of programming tasks are already being done by AI. xAI just cut 500 coders. Those two data points are connected. The question isn't whether AI will affect your role. It's whether the company you're joining sees you as someone who works alongside AI or someone AI will eventually replace. Lead with the judgment, relationships, and cross-functional work that compounds over time. That's your leverage.

If you've been searching for 6+ months, recalibrate your targets. The February jobs report showed 1.9 million people in long-term unemployment, up 27% from a year ago. If you're in that group, doing the same thing and hoping for different results isn't a plan. Look at whether your target companies are in sectors gaining capital or losing it. Look at whether the roles you're applying to are at companies showing growth signals or stress signals. Adjust your targeting, not just your resume.

Every week, we scan all active companies in the Kinship network for layoffs, funding rounds, financial trouble, and hiring freezes. This week: 1,568 companies checked. 112 signals found. When a company triggers a signal, it flows directly into your job scores. A strong role match at a company that just cut 40% of its workforce will score lower than the same match at a company with fresh funding and a growing team. You'll still see it if the fit is strong. But the risk is visible, not hidden.

You shouldn't have to piece together layoff trackers, earnings reports, and LinkedIn rumors to figure out if a company is safe to join. We do that scan every week so you don't have to. The signals are already in your scores before you apply.