139,000 Tech Workers Laid Off in 2026.

One thousand a day. Three hundred and twenty-five companies. The pattern is no longer about budgets, and senior engineers should stop reading it that way.

A thousand people. Every day. For five months.

That is the math of tech layoffs in 2026 so far. One hundred and thirty-nine thousand people across three hundred and twenty-five companies, tracked live by Trueup, updated daily, and the pace has not slowed once since January.

Oracle is the one everyone is talking about this week. Internal estimates from the May bloodbath put the cut between ten and thirty thousand. Oracle has not confirmed a public number. The Blind thread I read yesterday had one comment near the top that said “I have never seen such a big layoff in my life.” That commenter has been in tech for fifteen years.

But Oracle is not the headline. Atlassian cut ten percent in March while reporting record revenue. One point seven nine billion dollars, twenty-nine percent cloud growth, and they still cut. Walmart relocated a thousand corporate workers as it consolidated global tech and product teams. DeepL announced cuts of twenty-five percent in May. Cisco. Meta. LinkedIn. Roblox. Microsoft cut six thousand in April. The list runs to three hundred and twenty-five names.

If you are still reading this through the framework you used in 2023, you are missing the actual story.

That is what this piece is about.

The 2024 layoffs were the template. Nobody read the template.

Two years ago Google laid off engineers from the Flutter, Dart, and Python teams a few weeks before Google I/O. CNBC put the number at around two hundred. The Flutter community read between the lines for weeks. What looked like one company reorganizing one product team turned out to be a template that the rest of the industry is now running at scale.

Three structural signals from 2024 are repeating right now. Each one has a 2024 example and a 2026 equivalent. Read these in order.

Signal one. The “product, not priority” framing.

When Google reorganized in 2024, the Flutter layoffs were not about Flutter being unprofitable. They were about Flutter being classified as a maintained product rather than a strategic priority. A product gets shipped and supported. A priority gets staffed and defended. Google made Flutter a product. That is what the layoff actually said.

Atlassian did the same thing in March. Record revenue, twenty-nine percent cloud growth, and a ten percent reduction in the same window does not read as a budget signal. It reads as a reclassification. The people doing the work did not change. The category they sat in did.

Read this in your own company. Has your team been moved in the last six months from a priority to a product? The reclassification usually happens at the leadership level a quarter before the headcount move. Watch for org chart changes that put your team under a director whose mandate is “operational excellence” instead of “growth.” That is the marker.

Signal two. Senior engineers leave first, voluntarily.

Tim Sneath was the product lead for Flutter for years. He left for Apple in 2023, before the layoffs. Brandon DeRosier built Impeller, Flutter’s most important shipment in five years. He left for Google’s Android XR team in 2025. Neither was laid off. Both chose to leave.

The Atlassian story has the same shape. Vasilios Syrakis, a senior engineer who maintained their internal load balancer infrastructure for eight years, got cut in March. His response was to publish a forty-minute video walking through the entire Atlassian architecture stack. Public. Free. Anyone can copy it.

That response only makes sense if senior engineers were already deciding the company was no longer the place to do their best work. The layoff was the conclusion of a decision that had already been made by the people who matter. When senior people start leaving voluntarily, the layoffs are not a surprise. They are confirmation. The signal happens twelve to eighteen months before the cut.

Signal three. The gap between financial health and engineering investment.

Atlassian’s executive layer authorized two and a half billion dollars in stock buybacks and the CEOs sold a hundred and thirty-four million dollars in shares in the same window they announced the layoffs. Oracle, reportedly profitable, reportedly growing, is cutting between ten and thirty thousand people.

This gap did not exist a decade ago. Engineering used to get funded when the company was healthy. The relationship is more direct now. Your team gets staffed if the work it does cannot be done by an AI in eighteen months. Otherwise the work is on borrowed time.

That is the sentence nobody is writing. The 2026 layoffs are about who gets to keep their job once an AI does the work.

What to actually do this quarter

Skip the standard layoff advice. Update LinkedIn, network, side project, learn prompt engineering. You have read all of it. None of it is the actual move.

Four parts.

One. Document what you do that an AI cannot.

Not your job description. The specific judgment calls you made this year that required context an AI does not have access to. The architecture decision you pushed back on because you had seen the same pattern fail at a previous company. The bug you found because the latency profile matched something you debugged three years ago. The hire you advocated for because of a soft signal in the interview that the rubric did not capture.

Write these down. With dates and outcomes. Most senior engineers cannot articulate this and get caught flat-footed when their company asks them to.

Two. Stop investing in skills that have a clear AI replacement timeline.

If you spent the last six months getting faster at writing boilerplate components or shipping standard CRUD endpoints, you spent that time wrong. Those skills are getting cheaper every quarter. The market price for them is going to zero on a timeline measured in single-digit quarters.

The skills with longer half-lives are systems thinking, debugging at the boundary between systems you did not design, distributed-systems failure modes under realistic load, security work that requires reasoning about adversaries, and leading other engineers through complex tradeoffs. Spend your skill investment there. Not because they are immune. Because they degrade more slowly.

Three. Get one external thing that is yours.

A piece of writing. A side project with users. A talk someone remembers. A library you maintain. Something with your name on it that lives outside the company. This is the part of the Vasilios Syrakis story that most coverage missed. The reason his response landed is that the knowledge dump was something only he could publish. The company could take his salary. They could not take the eight years of context.

You do not need a viral hit. You need one thing that exists outside the org chart. The cost of building it now, while you are employed, is a few weekends. The cost of needing it and not having it is six months of job search starting from zero.

Four. Read the org chart, not the all-hands.

Companies tell you the strategy in the all-hands. They show you the actual strategy in the org chart. If your director’s title got changed from “VP of Engineering” to “VP of Engineering Operations” in the last six months, the strategy changed. The all-hands will not mention this. The org chart already did.

If three layers of management above you got compressed into two, the strategy changed. If your team’s quarterly objectives went from feature-driven to efficiency-driven, the strategy changed. If new hires are coming in at lower levels than the people who left, the strategy changed.

These signals are public inside the company. The engineers who read them are the ones who are not surprised when the layoff happens.

What this actually is

Not every layoff is a signal. Some are noise. Bad bets. Botched acquisitions. Single-quarter overspending getting clawed back. Those happen. Not the story.

The story is the structural shift in what engineering work is worth, on what timeline, to a company that has access to AI tooling that did not exist two years ago. The 2024 mobile layoffs were the first version of this story. The 2026 wave is the second version at fifty times the scale.

There is going to be a third version. Late 2026 or early 2027. It will hit harder because the AI tooling will be eighteen months further along. The companies cutting in 2026 are running the first iteration of an experiment. The ones cutting in 2027 will have the playbook.

The senior engineers who will be fine in 2027 are not the ones who learned the most prompt engineering tricks. They are the ones who can answer, in concrete and specific language, what they do that the AI cannot. The ones who cannot answer that question are not bad engineers. They are engineers whose value sits in the percentage of work that is getting automated.

If your reaction to this piece is “I do not know what I would say if my manager asked me that question on Monday,” you have your weekend assignment.

You do not need a five-year plan. You need a one-page document that says, specifically, what you do that an AI cannot, with examples from the last six months. Write it this weekend. Read it Monday before standup. Update it monthly.

That document is the thing that does not show up in the layoff numbers.

If you want more on how the 2024 layoffs actually played out and what kept Flutter shipping anyway, I wrote about the day ***Google laid off the Flutter team.***

I write about production mobile engineering with receipts. Follow if that is useful. ❤