If you work in tech, 2026 has been a nervous year. Headlines about AI replacing developers sit next to layoff announcements from companies that are still profitable. An earlier version of this post claimed the situation was “worse than 2008”. That’s not supported by the data, so this rewrite sticks to what the numbers actually show and what’s genuinely useful for developers to do about it.
The numbers
The most widely cited source on US layoffs is the monthly report from outplacement firm Challenger, Gray & Christmas. Here’s what it found for 2026 so far:
- First half of 2026: US employers announced 443,604 job cuts. Tech accounted for 139,156 of them, nearly a third, up 83% from 76,214 in the first half of 2025 (HR Dive).
- AI as a reason: 101,743 announced cuts in the first half were attributed to AI, 23% of the total. AI was the most-cited reason for several months running.
- Through July: tech cuts reached 149,023 for the year, and AI-attributed cuts 112,713. But the pace slowed sharply in the summer: July’s total was 33,429 cuts across all sectors, and Challenger noted that hiring plans picked up.
Andy Challenger summed up the first half: “Tech remains the epicenter of this year’s cuts. AI is the dominant force as companies are restructuring around it, automating roles, and reallocating budgets toward new capabilities.”
How to read these figures
A few caveats matter if you want an accurate picture:
- These are announcements, not confirmed job losses. Challenger counts announced plans. Some cuts happen gradually, some are offset by rehiring, and some never happen.
- “AI-attributed” is what companies say. Citing AI can mean real automation. It can also mean moving budget from salaries to AI infrastructure, or simply a more investor-friendly way to describe cost cuts.
- Tech is a specific slice. A surge in tech layoffs can happen while the wider job market is fairly stable.
Is it worse than 2008 or 2020?
Not in the way that phrase suggests. The 2008–09 recession and the 2020 pandemic shock caused millions of job losses across the whole economy. 2026 looks different: a sector-specific restructuring, heavily concentrated in tech and tied to a technology shift, rather than an economy-wide collapse.
For the people affected, that distinction doesn’t soften anything. But it matters for predicting what comes next. Recession-driven layoffs tend to reverse when demand returns. Restructuring layoffs change which jobs exist. Some roles come back in a different form, and some don’t come back at all.
Where the pressure is, and where the demand is
Based on company announcements and hiring patterns reported through 2026, the pressure is heaviest on:
- Entry-level and junior roles built around routine tasks that AI tools now speed up, such as simple CRUD features, boilerplate and basic testing.
- Support, operations and content roles where companies are deploying AI agents and chatbots.
- Middle-management layers that companies are flattening as part of wider restructuring.
Demand has held up better for people who can:
- build, deploy and evaluate AI-powered features in real products, not just call an API,
- own security, reliability and infrastructure, which get more important as systems get more automated,
- understand a business domain deeply, such as healthcare, finance or logistics, and translate it into software,
- review and take responsibility for code, including AI-generated code, at scale.
What developers can actually do
Generic advice like “upskill in AI” isn’t much help. Here’s more specific guidance.
1. Get good at working with AI tools, visibly
Hiring managers increasingly expect developers to use coding assistants well. That means knowing how to brief them, review their output and catch “almost right” code, not just having used them. Build something with an agentic workflow and write up how you kept it correct: tests, reviews, the mistakes you caught. That shows judgement, which is the scarce skill.
2. Build one real AI feature end to end
For example, a search feature over your own documents with embeddings, a support-ticket classifier with an evaluation set, or an internal tool that calls a model with proper error handling and cost limits. Put it on GitHub with a README explaining the trade-offs. Employers want evidence you can ship, measure and maintain AI features, not a certificate.
3. Go deeper in your stack
Routine work is easiest to automate. Deep knowledge isn’t. If you’re a frontend developer, that could mean performance, accessibility, state management at scale or framework internals. For backend developers, it could be databases, distributed systems or security. Depth is what lets you catch what AI tools get wrong.
4. Pair technical skills with domain knowledge
A developer who understands hospital workflows, logistics or payments is far harder to replace than one who only knows a framework. If you already work in a domain, lean into it: learn its rules, regulations and data.
5. Make your work easy to verify
With so many applicants, a clear portfolio helps: a few focused projects with real READMEs, technical blog posts explaining problems you solved, and contributions to open source. Show your reasoning, not just finished code.
6. Look beyond big tech
Much of the cutting is concentrated at large tech companies. Smaller companies, non-tech businesses going through digital transformation, government contractors and regulated industries often still struggle to hire experienced developers.
7. Protect your finances and network early
Keep an emergency fund, keep your CV and LinkedIn up to date even when you’re happily employed, and stay in touch with former colleagues. Most jobs still come through people, not job boards.
The outlook
The honest answer is that nobody knows how far this restructuring goes. The summer slowdown in announced cuts and the pickup in hiring plans are encouraging, but the direction is clear: companies are reorganising around AI, and the mix of skills they want is changing.
For developers, the most reliable position is the one that’s always been reliable, just more so now: understand problems deeply, take responsibility for correctness, and use every tool, including AI, to deliver more than you could alone.
Related reading
- Is AI Hurting Open Source?
- A Practical Workflow for Writing Software with LLMs
- .com, .dev, .io or .online? Choosing a Domain Extension
Sources: HR Dive on the Challenger H1 2026 report, Challenger, Gray & Christmas July 2026 report.
