Picture this. You graduated top of your class in the early 2000s, got a well-paid job at a company everybody's heard of, and spent twenty years grinding out a career. On March 31st, 2026, you wake up at 6 a.m. and find an email from corporate telling you your position has been eliminated, effective immediately. No meeting. No warning. Just a link to HR and a box waiting for you at the front desk.
The Oracle story is the one that made headlines, but if you've been paying attention for the last two years, you've seen this movie at Google, Meta, Microsoft, and Amazon. Layoffs.fyi has been tracking the carnage since 2020 and the cumulative number is now north of half a million tech jobs. People are angry. They should be.
But here's the thing nobody seems to want to say out loud. A huge part of this mess is a hangover from 2021, and the other huge part is a story about what happens when supply explodes faster than demand. It's brutal, but it's not the end of software engineering as a profession. I'll explain why in a minute. First, let's be honest about how we got here.
Last updated: April 2026
The 2021 Overhire Everyone Forgot About
Cast your mind back to 2021. The pandemic had moved everything online almost overnight. Zoom was a verb. Shopify stores were minting millionaires. Every B2B SaaS with a pulse was doing 150% year-over-year growth, and every company with a tech budget decided they needed to triple their engineering headcount before their competitor did.
The numbers are wild in retrospect. Meta went from roughly 58,000 employees at the end of 2020 to over 87,000 by late 2022. Amazon's workforce doubled from 798,000 to 1.6 million over the same stretch. Google hired over 30,000 people in 2022 alone. None of this was happening because the companies were running out of engineers. It was happening because the companies were terrified they'd look understaffed compared to the other companies panic-hiring next door.
Then 2022 ended. Interest rates climbed. The post-pandemic demand normalized. Leadership looked at the headcount spreadsheet, did some basic arithmetic, and realized they'd hired for a world that wasn't going to exist. The first wave of layoffs in early 2023 was framed as "course correction." The second wave in 2024 was framed as "efficiency." By 2026 nobody even pretends anymore — it's just the thing that happens every few quarters.
So, yes. The layoffs are real. They're brutal. Losing your job is never okay, and I don't want to diminish what hundreds of thousands of people are going through. But some of this is companies cleaning up after their own mistakes, and the engineers paying for it didn't make those mistakes. You didn't over-hire yourself.
The Junior Developer Trap
If the layoffs are bad for senior engineers, the market for new grads is apocalyptic. Computer science was the degree of the 2010s. Every parent, every high school counselor, every think piece told kids the same thing: "just learn to code, the jobs are always there." CS enrollment roughly quadrupled between 2010 and 2022, and those graduates are now hitting the job market at the exact moment entry-level roles have collapsed.
The math is brutal. A viral LinkedIn post from late 2024 documented a single junior developer posting that received 400 applications in under 2 minutes. Not two days. Two minutes. Friends of mine have built Python scripts that spray-apply to every "entry level software engineer" posting on LinkedIn until they hit the platform's daily rate limit. After a week and thousands of applications, they might get two replies. Usually both are rejections.
This isn't about talent. The talent is there. Supply just ran way past what the market wanted, and companies are using that imbalance to be selective in ways that would have been unthinkable five years ago. A role that would have been filled by a bright new grad in 2019 now requires three years of experience, two internships at name-brand companies, and a portfolio of shipped side projects. The phrase "entry-level" has quietly stopped meaning anything.
The Computing Research Association's annual Taulbee Survey has been tracking CS enrollment for decades and the numbers paint a clear picture: we trained too many engineers for a market that also decided, at the same time, it wanted fewer of them. Bad timing is a polite way to put it.
AI and the "Vibe Coding" Hype
The other story on top of all this is, of course, AI. Every CFO has now read the same three articles suggesting that one mid-level engineer with Cursor and Claude Code can replace three junior developers. On paper, this looks like a cost optimization. In practice, the evidence is messier.
The GitClear 2024 study on AI-assisted code found that AI-generated code is significantly increasing what they call "code churn" — the rate at which new code gets rewritten shortly after it's committed. The code gets written fast. Then it gets rewritten just as fast because it breaks, doesn't hold up, or doesn't actually solve the problem. A follow-up 2025 study from Uplevel found that AI coding tools didn't meaningfully increase throughput and slightly increased bug rates for teams using them heavily. DORA's 2024 State of DevOps report reached a similar conclusion: AI tools help individual productivity metrics while slightly hurting team delivery performance.
None of this is an argument that AI tools are useless. They're not. I use them every day. The argument is that the "replace three juniors with one mid-level + AI" math is mostly a fantasy running on vibes, and the people running the spreadsheet never factor in the cost of the bugs, the rewrites, and the institutional knowledge that quietly evaporates when you stop hiring the people who were supposed to become your next senior engineers.
But try explaining that to a CFO whose only job is to cut costs before the next earnings call.
The EA Problem (Or: Who Actually Gets Paid When Things Go Well)
Here's the bit that makes people the angriest, and rightfully so. Electronic Arts reported over $7.4 billion in revenue in fiscal 2024. They also laid off roughly 5% of their workforce that same year. The CEO's total compensation for the year was publicly disclosed in their proxy filing at around $25 million.
You ship a game people love, and the reward for the team that built it is a calendar invite and a box to pack your stuff in. Meanwhile, the executives who made the strategic calls that led to over-hiring and then under-shipping take home more money than most of the laid-off employees will earn in their entire careers.
This pattern isn't unique to EA. Look at any of the big layoff rounds and you'll find the same story: record revenue, record executive comp, and rank-and-file employees carrying the cost of decisions they didn't make. Harvard Business Review did a deep dive on the pattern back in 2023 and the conclusion was pretty damning — most layoffs don't actually improve long-term financial performance, but they reliably juice the next quarter's numbers, which is the only thing the comp committee cares about.
So when someone tells you layoffs are "just business," remember that the business math almost never works. It's theater.
The Interview Process Is Broken
Let's talk about the thing that every actively-job-hunting engineer I know is completely burned out by. The process itself.
You find a posting. You apply. You wait two weeks. You get a recruiter screen. You wait another week. You do a coding screen. You wait. You get a full technical round. You get a take-home project that eats a full weekend. You wait. You do a culture-fit panel. You wait. Three weeks later you get a form rejection, or worse, you hear nothing at all.
And at the end of all that, the salary waiting for you is lower than the same role paid in 2021, because the supply-side math has shifted and companies know they can offer less.
On top of that, a lot of the jobs aren't even real. "Ghost jobs" — postings that companies keep up for tax reasons, pipeline building, or just to make themselves look like they're growing — have become common enough that Resume Builder's 2024 survey found 40% of companies admitted to posting fake listings in the last year. Two in five. The posting you just spent a weekend preparing for might not correspond to a real job.
If you feel like the process is gaslighting you, it's because in many cases it literally is.
We've Been Here Before
Now for the pushback, because I don't want to leave you in the pit.
The dot-com bubble bursting in 2000 was genuinely worse in some ways than what we're living through now. Serious people at serious publications wrote confident obituaries for the entire internet. Pets.com collapsed. Webvan collapsed. The NASDAQ lost 78% of its value between March 2000 and October 2002. Engineers who had been making fortunes in 1999 were delivering pizzas in 2002. Entire categories of companies — online grocery, online pet supplies, streaming video — were declared permanently unviable.
And then, slowly, the companies with actual substance kept building. Amazon survived. Google went public in 2004 and changed how the internet worked. The technologies we use every day in 2026 — React, TypeScript, cloud infrastructure, the App Store, modern search — all came out of what people in 2002 were calling "the wreckage." The bubble popping didn't kill software. It burned away the nonsense and left the real work behind.
I think that's what's happening with AI right now. We're in the hype phase. Every company is rushing to look like they're ahead of it. Shareholders are impressed when a CEO talks about "AI-first" and "replacing engineers with agents." Gartner's 2025 predictions have put generative AI at the "peak of inflated expectations" and are forecasting the usual slide into the trough. The results from the replace-engineers-with-AI strategy are inconsistent at best and quietly disastrous at worst. At some point, the industry will have to get honest about what AI is actually good at and what still needs a human being making a judgment call.
And there's one thing no current model can do, no matter how capable it gets: it cannot have a bad idea at 2 a.m. that turns out to be brilliant by morning. It cannot feel a user's frustration and rethink a product from the ground up. It cannot take ownership of a decision that might be wrong. Anthropic's own engineering team wrote a careful essay on this where they lay out exactly which parts of agent work still need human judgment, and the list is longer than the AI maximalists want to admit.
So Where Does That Leave Us
Somewhere uncomfortable. For now.
The industry over-hired in 2021, panicked in 2023, and then swerved hard in the other direction. A generation of CS grads is hitting the worst entry-level market in twenty years. AI is real and useful and not going away, and the vibe-coding crowd is going to keep claiming it has replaced engineers until the bugs they've shipped come due. None of this is fair. Most of it isn't even rational from a business standpoint. It's just what happens when everybody panics at the same time.
But here's what I'd tell anyone on the receiving end of this: the developers who understand why they're building what they're building — the ones who can hold a problem in their head, feel a user's frustration, and take ownership of a decision — are not the ones getting replaced. They're the ones getting hired at the places that have stopped panicking and started trying to build something that actually works.
The way to become that kind of developer isn't to grind LeetCode or spray-apply to 1,000 listings. It's to build things. Real things. Things that have opinions and tradeoffs and a point of view. Patrick McKenzie has been writing about this for years — the most underrated skill in this industry isn't coding, it's the ability to explain a problem, design a solution, and ship it to the people who actually care. AI isn't taking that job. Nothing is.
If you're laid off right now, or can't get your first job, or are watching the ground shift under your feet, I'm sorry. It's not your fault. It's also not the end. The industry has survived worse, and the people who build the next thing on top of this mess are going to come from the same group of humbled, scarred, still-curious engineers who lived through it.
Build the interesting things. Ship them. The rest will sort itself out eventually. It always does.
If you're one of those developers building something interesting in the AI / conversational space, Chatsby is free to try and a decent way to turn a side project into something users can actually talk to. No pitch beyond that — good luck out there.
