45,700 Tech Workers Cut in 2 Months. AI Is the Reason Given. Here Is Which Companies, Which Roles, and What Comes Next.

By the end of February 2026, 45,700 technology workers had been laid off across the industry — in just two months. The pace is running above 2024 levels, which were themselves elevated above the pre-AI-era baseline. What is new in 2026 is the language companies use to explain the cuts: AI efficiency, automation, productivity gains, and the ability to do more with smaller headcounts.

This is not a recession story. Most of the companies cutting workers are profitable and growing revenue. It is a restructuring story — and the restructuring is accelerating because AI tools have genuinely changed what a team of five engineers can deliver compared to what required a team of twenty two years ago.

Here is the data, the pattern, and what to actually do if you are in or adjacent to the affected categories.

The Numbers: Who Cut and How Many

Google (Alphabet): Ongoing cuts throughout Q1 2026 across multiple divisions. Google has cited AI automation of tasks previously requiring large engineering and operations teams. The cuts are concentrated in middle management layers, developer relations, and repetitive engineering functions. Google's AI investments are growing while its total headcount shrinks.

Microsoft: Announced further reductions following the OpenAI investment doubling in January 2026. Microsoft has been explicit that AI Copilot tools have reduced the need for certain categories of support, QA, and documentation roles. Approximately 1,000-1,500 positions in the Q1 2026 round, with more reported in pipeline.

Meta: Cut approximately 3,600 workers in February 2026 — described as performance-based but with AI efficiency as the stated structural reason for not backfilling. Meta's AI investments in Llama 4 and internal tooling have allowed the company to reduce engineering headcount while increasing output.

Salesforce: CEO Marc Benioff has been the most explicit of any major tech leader: "AI means we need dramatically fewer people to do the same work." Salesforce has cut aggressively in sales, support, and engineering roles traditionally associated with CRM implementation and customisation. These are tasks that AI agents now handle in part.

Amazon: AWS-related headcount reductions in Q1 2026, particularly in the middle layers of technical program management and solution architecture — roles that required human coordination which AI tools now partially automate.

Intel: Continuing its multi-year restructuring under Pat Gelsinger's successor, with manufacturing and engineering roles reduced as Intel adapts to a world where GPU and custom silicon (not general-purpose CPUs) capture the high-margin AI workload.

Smaller companies: The Computerworld tracker shows 45,700+ total, which means the cuts are spread across dozens of companies beyond the household names. Fintech, edtech, SaaS, and mid-stage startups are cutting heavily — many of these companies raised large rounds in 2021-2022 and are now right-sizing to sustainable unit economics while discovering that AI reduces the headcount required to maintain the same product.

The Role Pattern: What Is Being Cut

QA and testing roles: AI-assisted testing tools (Copilot for testing, GitHub Actions with AI test generation, purpose-built QA AI) have dramatically reduced the human labour required for manual testing, regression testing, and test case generation. QA was already a shrinking role; AI tools have accelerated the contraction.

Technical writing and documentation: AI tools write documentation effectively from code and prompts. Teams that employed 3-5 technical writers for product documentation are discovering one person with AI tools produces comparable output.

Tier 1 and Tier 2 support: AI agents handle the majority of routine technical support queries. Companies that ran large support operations find the human support headcount required is now a fraction of what it was in 2022.

Middle management in engineering: When AI tools reduce the coordination overhead of engineering work — generating summaries, tracking decisions, surfacing blockers — the management layer required to coordinate teams shrinks. This is not universal, but "remove a layer of management" is now a viable efficiency move in ways it was not before AI tools.

Data entry, reporting, and analyst roles: Roles that primarily involved moving data between systems, generating standard reports, or summarising information are being eliminated at scale. These were already targets for automation; AI has made the automation effective enough to act on.

Roles involving content generation at volume: SEO writing, marketing content at scale, email campaign copywriting — anywhere content generation was a high-volume labour task is being reduced.

What Is NOT Being Cut (or Is Growing)

AI/ML engineering: Demand for engineers who can build, fine-tune, evaluate, and deploy AI systems is high and growing. The engineers building the tools that eliminate other roles are not themselves being eliminated.

Security engineering: The attack surface created by AI systems, AI-generated code, and the broader AI-enabled threat landscape requires more security engineers, not fewer. Every company that deploys AI features creates new vulnerabilities that need security expertise.

Prompt engineering and AI product management: Understanding how to make AI systems produce reliable, useful output in production has created a new specialist role. This is a rapidly evolving domain.

DevOps and platform engineering: The infrastructure required to run AI at scale — GPU clusters, vector databases, model serving, monitoring — requires engineering expertise that is in short supply.

Engineering in regulated domains: Healthcare AI, fintech AI, legal AI — domains where AI output requires human validation and where regulatory compliance is complex retain high demand for engineers who understand both the domain and the AI.

Senior engineers who pair AI with deep expertise: The engineers whose value compounds when combined with AI tools — who use AI to multiply what they can accomplish rather than as a replacement — are more valuable than before. The engineers whose value was primarily volume of code output rather than quality of decisions are more exposed.

The Honest Structural Picture

AI tools in 2026 have genuinely changed the production function for software development. Tasks that required a team of twenty can now be accomplished by a team of eight. This is not a prediction — it is what companies are reporting in their public disclosures and earnings calls.

The companies cutting fastest are not cutting because they are struggling. Google, Meta, and Microsoft are profitable and revenue-growing. They are cutting because they have found that AI tools have created genuine capacity for reduction without production impact. CEOs are now being asked by boards why they maintained the 2021 headcount when AI tools have changed the math.

This dynamic will continue. The productivity improvements from AI coding tools, AI-assisted testing, AI documentation, and AI support are real. The headcount reductions follow the productivity improvements with a lag of 6-18 months as contracts expire, performance review cycles complete, and reorganisations are planned.

The Geography of the Cuts

The cuts are concentrated in roles that can be performed remotely with AI assistance — which means they are not geographically isolated to Silicon Valley. Tech workers in India, Eastern Europe, Southeast Asia, and Latin America who were hired during the 2020-2022 expansion into remote-friendly roles are affected alongside US workers.

India specifically: the large engineering services and outsourcing sector that grew significantly during the remote-work era faces structural pressure as companies find that AI tools reduce the benefit of labour arbitrage. A US-based engineer with AI tools can produce more output than a team using older workflows, closing the cost gap that drove offshore hiring.

This is one reason your site's India traffic (31% of total) is so high — Indian tech professionals are intensely interested in the AI displacement question. They are directly affected.

What To Do If You Are Affected or At Risk

If you have been laid off:

The job market for engineers who can effectively work with AI tools is better than for engineers who cannot. The fastest path to re-employment is demonstrating AI-augmented capability, not just traditional engineering skills. Your portfolio should include projects built with AI assistance — the goal is to show you multiply your output with AI, not that you resist it.

The strongest resume signal in 2026: shipped a production feature using AI tools (Cursor, Copilot, Claude Code) that would have taken 3x as long without them. Concrete, demonstrable productivity gain.

If you are currently employed and concerned:

Audit your work honestly. Which parts of your current role are high-volume, repetitive, and could be done by AI with a human reviewer? If a significant fraction of your time is in that category, that is the vulnerable surface area. The response is to shift your time and skill development toward the parts of your role that require judgment, domain expertise, system design, and cross-functional influence — the areas where AI augments rather than replaces.

The hardest category: mid-career engineers in platform/tools roles:

Engineers who built their expertise on a specific toolchain (testing frameworks, CI/CD configuration, documentation systems) and whose primary value was deep knowledge of those tools are most exposed. The tools are being automated. The skill to retrain on: evaluating, debugging, and improving AI-generated output in your domain. This is different from using AI tools; it is understanding when AI output is wrong and why.

Learning investments with the best return in 2026:

• Prompt engineering and AI evaluation: Understanding how to make LLMs produce reliable output in production contexts
• AI security: Prompt injection, model extraction, data poisoning — the security problems specific to AI systems
• MLOps and AI infrastructure: Running models in production at scale
• Domain depth: In any regulated or complex domain, deep expertise remains valuable because AI tools require expert oversight to be useful safely

The Next 18 Months

The 45,700 cuts in January-February 2026 are not the peak. The cuts from Q1 2026 reflect decisions made in Q3-Q4 2025 as companies assessed their AI tool deployments and planned reductions. The decisions being made now — as Llama 4, Claude 3.7, and GPT-5 are deployed across enterprise tooling — will produce cuts visible in Q3-Q4 2026.

The companies that will hire robustly in 2026: AI-native startups that are building new products (not replacing existing ones), established companies building AI infrastructure, and regulated sectors deploying AI for the first time and needing engineers who understand both domain and AI.

The long-term outlook for software engineering is not elimination but transformation. The number of engineers required to maintain a given system is declining. The complexity of systems that can be built is increasing. The net effect on total employment is genuinely uncertain — the "automation creates more jobs than it destroys" argument has historical support but also historical exceptions. What is certain: the engineers who adapt to AI-augmented work will fare significantly better than those who do not.

---

45,700 jobs in two months. AI cited as the structural driver at every major company. The honest question for every developer is not "will AI affect my job?" — it is "how much of my current role is vulnerable and what am I doing about it?" The window for a deliberate, calm response to that question is narrowing.