In the first seven months of 2026, more than 165,000 roles were cut across the corporate world. September is bringing another wave. Oracle, Pinterest, Samsung and eBay have all scheduled workforce reductions this month.
The numbers are grim, but they are not new. Layoffs happen in every cycle.
What is new is the language executives are using to explain them.
“Lean, fast, and AI-native”
That phrase came from Coinbase CEO Brian Armstrong, who used it to describe how the company was restructuring itself. He has also talked about testing very small teams that manage AI agents instead of larger human teams.
CrowdStrike CEO George Kurtz filed a memo with the SEC saying AI flattens the company’s hiring curve and speeds up the path from idea to product.
Accenture set the template earlier. In late 2025 it announced roughly 11,000 job cuts as part of an $865 million restructuring. CEO Julie Sweet told investors the firm would move staff out on a tight timeline where retraining was not realistic.
Read those together and a pattern shows up. None of these are described as a hiring pause. They are described as a redesign.
The job market is splitting, not shrinking
Here is where it gets more complicated than “AI takes jobs.”
In a Resume.org survey, 47% of companies said they are hiring more technical or AI-focused staff in 2026. Another 48% said they are hiring more people who can use AI tools well.
So headcount is falling and hiring is climbing at the same time. Just not in the same roles.
Reporting on 2026 tech layoffs also notes that the cuts have reached senior positions and specialised technical jobs. This is not only entry-level work disappearing. Restructuring is going deeper into the org chart than it usually does.
Look at which jobs are going
The pattern in the actual cuts is worth studying.
Paycom let go of more than 500 people after automating payroll and back-office functions. Google Cloud eliminated over 100 design and user-experience roles while moving money toward AI engineering. GitLab cut around 14% of its staff and said the savings would go into AI products.
Marketing, operations, support, design, payroll, back office. Roles built around routine processes that follow the same steps every time.
That is the common thread. Not “junior” work. Repeatable work.
The case against all of this
Not everyone thinks this ends well, and the counterargument is worth understanding even if you disagree with it.
Bloomberg ran an opinion piece by Gautam Mukunda arguing that AI-fuelled layoffs will backfire. The core point is that cutting staff based on what a technology might do is a bet, not a plan. If the tools underdeliver, you have lost institutional knowledge you cannot simply buy back. Rehiring is slow and expensive. Companies made a version of this mistake during pandemic-era over-hiring, then again during the correction.
There is also a measurement problem.
“AI-driven” is a convenient label. A company cutting costs because growth slowed can attach AI to the announcement and sound forward-looking instead of struggling. Oracle’s annual SEC filing showed 21,000 workers let go under AI-related restructuring. Whether AI caused those cuts or simply explained them is very hard to verify from outside the company.
Both things can be true at once. Some of these cuts are genuine redesigns. Some are ordinary cost-cutting wearing a better jacket.
What this means if you manage people
Three practical takeaways.
The valuable skill is not tool knowledge. Knowing how to prompt a model is becoming common. Being able to look at a process and redesign it so tools actually fit is not. Tool training is cheap. Process judgment is not.
Information-shuffling work is exposed now, not later. If most of your team’s day is moving data between systems, updating records and formatting reports, that work is directly in the path of what current agents do. Not in five years. This year.
If you are the one cutting, be straight internally about the bet. Teams can tell the difference between “we are restructuring around new tools” and “we are behind on revenue.” Getting caught dressing up the second as the first costs more trust than the headcount saves.
The strange upside
One thing has genuinely improved in this cycle.
Restructuring used to hide behind soft words. Focus. Alignment. Streamlining. Now executives write the reasoning into SEC filings and shareholder letters in plain language.
That openness is unpleasant to read if you are on the wrong end of it. But it is easier to plan around a stated reason than a vague one. If a CEO says the company is rebuilding around agents, employees know what to learn and what to stop investing in.
Whether the reasoning turns out to be correct is a separate question entirely. Nobody will be able to answer that for another two or three years.
