February 2025 was brutal. Across Meta, Microsoft, Amazon, Salesforce, IBM, and a dozen other major tech firms, 16,000 jobs disappeared in a single month. The headlines blared that “AI” was responsible, and the implication was clear: if your job involved anything a large language model could theoretically do, you were at risk.

Here’s the problem with that narrative: it’s both true and wildly misleading at the same time. Yes, AI was explicitly cited for approximately 4,680 of those layoffs. But which jobs, why, and whether those reasons have anything to do with what your role actually is are three very different questions that the press rarely bothered to separate.

Let’s untangle what actually happened in February and what it says about which roles are genuinely threatened versus which ones are actually growing.

The Numbers, Plainly

Meta cut 3,600 roles. The company explicitly blamed efficiency improvements and an AI-first strategy. Those cuts fell heavily into recruiting, finance administration, and mid-level management, with the theory that AI can screen candidates, process expense reports, and handle some of the coordination that previously required human judgment.

Microsoft cut 2,280 people, primarily from sales and customer service roles. The stated reason: integrating Copilot more aggressively into workflows meant fewer people needed to handle routine customer questions, and the sales organization could be flattened.

Amazon cut 2,100 roles. The company has been slower to blame AI specifically, but the timing aligns with their aggressive push on AWS generative AI services and their internal shift to reduce customer service headcount through automated systems.

Salesforce cut 1,000 and then later disclosed another 1,000. The CEO explicitly said some roles could be eliminated through AI automation.

IBM, notably, cut 8,000 administrative and back-office roles but simultaneously announced new hiring for engineers and technicians. This is the detail most coverage glossed over: IBM didn’t cut people because AI made tech obsolete. IBM cut people whose primary job was administrative overhead.

Add them up: about 4,680 of the 16,000 cuts that month came from roles that fall into a few clear categories: recruiting support, finance operations, customer service agents, sales support, and administrative overhead. The common thread isn’t “jobs AI can do better.” It’s “jobs where AI could handle the routine part of the work, freeing up or eliminating the humans who were there to handle the routine part.”

That’s a meaningful distinction and it changes what you should be worried about.

The Roles Actually Under Pressure

If your job has significant repetitive, rule-based components, you should take this seriously. That doesn’t mean you need to panic. But it means you should think clearly about what part of your role is defensible and what part isn’t.

A customer service agent who handles mostly straightforward questions about order status or password resets? That role is genuinely under pressure. An AI chatbot can handle those interactions at scale and people who were mostly managing routine issues are seeing their positions disappear. Some companies are retraining those people into more complex support roles, but that requires the company to care about retention and to have more complex work to assign.

A recruiting coordinator whose job is screening resumes and scheduling interviews? That role is exposed. Resume screening is exactly the kind of task an AI system can reduce from a day’s work to an hour of human review of the shortlist. It doesn’t mean recruiting becomes fully automated, but it means you don’t need as many people doing the repetitive parts.

A finance operations person whose primary work is processing expense reports and reconciling statements against rules? That’s a role under genuine pressure from automation, both AI and RPA systems that predate LLMs.

The pattern isn’t that any particular job is going away entirely. It’s that the routine parts are being compressed, which means you need fewer people or you need people who can do the complex parts that AI can’t.

The Roles Actually Growing

Meanwhile, at the same companies doing the cutting, certain teams are hiring. IBM hired engineers. Microsoft is hiring for AI product development and infrastructure. Amazon is bringing people into AWS AI services. Meta is hiring infrastructure engineers.

This is worth dwelling on: you can’t cut 3,600 people and simultaneously say the company is dying. The company is restructuring. It’s moving people and money toward the parts that make AI products work and away from the parts that manage the overhead of having too many people.

For IT work specifically, the growth areas are clear: infrastructure, platform engineering, and the specialized roles that make AI systems deployable at scale. Those roles require people who understand systems, who can think about reliability and cost, who understand how to make something work when it’s not a simple integration of a vendor API.

The second growth area is anything that requires understanding your business. An AI system that could theoretically do your job isn’t going to do your job well unless someone understands what the job actually needs to accomplish. That’s why companies are cutting support roles but keeping or hiring people who work with customers to understand what they actually need. The AI is a tool those people use, not a replacement for the thinking.

What This Means for Your Career Strategy

If you’re in a role that’s primarily routine work with some complexity layered on top, you should invest in becoming the person who handles the complexity. That might mean learning to use AI tools better than most people, so you become the person who makes sure the AI output is correct. It might mean pushing into the parts of the job that require judgment, context, and understanding, and letting someone else (or a bot) handle the routine parts.

If you’re in infrastructure or platform engineering, the fact that there are layoffs in other parts of the industry shouldn’t worry you much. The companies restructuring are moving money toward your function. The demand for people who understand how to operate complex systems at scale is growing, not shrinking.

If you’re in a role that’s currently 80 percent routine tasks and 20 percent complex work, you have two paths. You can try to become the person who owns the 20 percent and lets automation handle the routine part. Or you can learn to do something different. But the path of “I’ll just do the routine work better” is going away.

The harder truth is this: February 2025 wasn’t a normal layoff cycle. It wasn’t primarily driven by a down market or reduced revenue. It was companies making a strategic bet that they could do more with fewer people by using AI to compress the routine parts of work. That’s a one-time restructuring, not a permanent contraction. Which means there’s hiring happening in parallel with the cuts, but it’s hiring for different roles than the ones being eliminated.

IBM’s Model Is Instructive

IBM cut 8,000 back-office roles while hiring engineers. The subtext is clear: the company needs fewer people administering systems and more people building systems that administer themselves. That’s the broader pattern in tech right now. Companies are cutting overhead and reinvesting in people who make the systems work better.

If you’re reading this and thinking “I’m not an engineer,” take a moment to think about what that role actually is. An engineer is someone who solves problems with systems. That might be code, but it might also be understanding how a process works and making it work better. The company needs people who can think systemically and who can handle complexity. Those are the roles that survive and grow.

The reference to IBM cutting 15,000 and Cisco cutting 6,000 shows that massive restructuring isn’t new. But what’s different in 2025 is the direction of the cuts. In 2023 and 2024, companies cut to adjust to a slowing market. In February 2025, companies are cutting to redirect toward AI. The destination matters for what you should be learning and where you should be looking.

The Uncomfortable Reality Check

If February 2025 taught us anything, it’s that the safe play isn’t to do your current job better. It’s to become indispensable in a role that’s growing rather than to optimize your way into irrelevance in a role that’s shrinking.

That’s a harder piece of advice than “learn AI” or “upskill.” It might mean moving teams, taking a different kind of role, or learning things that don’t feel immediately connected to what you’re doing now. But it’s the honest assessment of where the job market is.

The 16,000 people laid off in February aren’t unemployable. Most of them will find jobs in a market that’s still hiring. But they will find jobs in a market that’s fundamentally restructured away from routine administrative work and toward people who can handle systems complexity.

The AI citations were real. But they weren’t about AI taking over thinking jobs. They were about AI handling the routine parts of jobs so that companies could reduce headcount in positions that were primarily routine. If your current role is primarily routine, the answer isn’t to get better at routine work. It’s to move toward roles that are primarily complex and use AI as a tool to handle the routine part more efficiently.

Companies like Salesforce and others are now looking at how layoffs actually play out. The pattern of cutting overhead while investing in core capabilities is likely to continue. Understanding which category your role falls into isn’t optional anymore. It’s the foundation for a rational career strategy in 2025.