Everyone's Quitting Computer Science — Here's Why I'm Not

"Computer science majors are disappearing," the Washington Post declared two weeks ago. The data backs it up: CS enrollment dropped 11.2% this year, the steepest decline of any major. Graduate enrollment fell 14%. For the first time in two decades, the University of California system saw CS enrollment decline across the board. TechCrunch is calling it "the great computer science exodus."

I'm a third-year CS student in Ontario. A year ago, my program was overflowing — waitlists for upper-year courses, TAs stretched thin, labs at capacity. This semester, I noticed more open seats. A few friends quietly switched to data science or mechanical engineering. And I'll be honest: for about a week in February, I wondered if they knew something I didn't.

They don't. They're panicking. And I think they're going to regret it.

The numbers are sending contradictory signals

Here's the thing nobody is putting side by side. CS enrollment is plummeting — and CS starting salaries are rising. According to NACE's 2026 Winter Salary Survey, the average starting salary for CS graduates hit $81,535 — up 6.9% from last year, the largest projected increase of any major. Hiring is roughly flat, yes. But flat hiring plus fewer graduates equals better odds for the graduates who remain.

Meanwhile, Handshake's Class of 2026 data shows 70% of CS seniors feel pessimistic about their career prospects. Only 14% are optimistic. That's the highest pessimism rate of any major surveyed.

Seventy percent pessimism. Seven percent salary growth. If this were a stock, every contrarian investor on earth would be loading up.

Where are the exodus students going?

They're not abandoning tech entirely — they're scattering. Mechanical engineering enrollment is up roughly 11%. Electrical engineering is up about 14%. New AI-specific degree programs are launching at Columbia, USC, and Northwestern this fall. The concern is that students are trading a proven CS degree for either a field with its own job market challenges, or a brand-new AI program with zero alumni network and untested curricula. Switching to a major that's never produced a graduate because you're scared of a major that's produced millionaires for three decades is a bold move.

We've seen this exact movie before

After the dot-com bust, CS enrollment dropped 32% between 2000 and 2004. Students fled the major like it was radioactive. The narrative was identical: "tech is over," "there are no jobs," "the bubble popped and it's never coming back."

By 2005, employment prospects for CS graduates were better than for students in any other discipline. The students who stuck it out through the trough graduated into one of the best job markets in the history of the profession. The ones who switched to business or communications because Pets.com went bankrupt? They missed the rise of Google, Facebook, AWS, and the entire mobile revolution.

The parallel isn't perfect — AI is a different kind of disruption than a speculative bubble popping. But the behavioral pattern is identical: students make career decisions based on headlines, not fundamentals. And by the time the headlines turn positive again, the window of reduced competition has closed.

The enrollment lag is a documented phenomenon

Student interest in CS trails the actual job market by 2-3 years. After the dot-com crash, tech employment recovered by 2003-2004, but enrollment didn't bottom out until 2007. Students who enrolled during the trough graduated into a boom. Students who left during the trough missed it entirely. The Computing Research Association's 2025 survey found that 62% of computing programs are already seeing enrollment decline — and if history rhymes, the students abandoning CS right now are making the same timing mistake their predecessors made twenty years ago.

The companies that said "no juniors" are changing their minds

Here's what I find most telling about the current moment. While students are fleeing CS, the companies at the center of the AI revolution are quietly opening their doors to entry-level talent.

Netflix — a company famous for only hiring senior engineers — now runs a formal New Grad program. OpenAI launched an Emerging Talent track designed for people with 0-3 years of experience. Anthropic explicitly doesn't require PhDs — roughly half their technical staff doesn't have one — and they actively recruit entry-level engineers.

These aren't charitable gestures. These companies are hiring juniors because they need them. AI systems require humans to evaluate outputs, build tooling, run safety evals, and do the kind of careful, methodical engineering work that doesn't require twenty years of experience but does require solid CS fundamentals.

And it's not just AI labs. Duolingo's CEO walked back the company's "AI-first" employee evaluation policy this month, telling staff "I'm not going to force you." Klarna — the poster child for replacing humans with AI — quietly reversed course after customer satisfaction tanked and started rebuilding its human workforce. The companies that leaned hardest into "AI replaces everything" are the ones learning, in public, that it doesn't.

I wrote about the trough of disillusionment a few weeks ago — we're living in it right now. On the other side of every trough, there's a lot of work to do. That work needs people.

What the pessimism is actually telling you

When I first saw that Handshake number — 70% of CS seniors are pessimistic — my reaction was "yeah, fair, it's rough out there." My second reaction was: wait. When has widespread pessimism in a market with rising prices ever been the right signal to follow?

I don't want to be flippant about this. The job market is harder than it was in 2021. Application volumes are up. Response rates are down. "Entry-level" postings still demand 2+ years of experience. I've written about this honestly, and if you're in the thick of it, the pessimism isn't irrational. It's a response to a real experience.

But the aggregate data tells a different story than the individual experience. Salaries are up. AI companies are hiring juniors for the first time. The total number of CS graduates competing for those roles is about to shrink significantly. CNN put it bluntly: "the demise of software engineering jobs has been greatly exaggerated."

If you're one of the students who stays, you're about to benefit from reduced competition in ways that won't be obvious for another 12-18 months. The students switching to mechanical engineering or a brand-new AI major aren't wrong to hedge — but they might be wrong about the timing.

What I'm actually doing about it

I'm not writing this from smug certainty. I had my own moment of doubt earlier this semester. I watched friends switch majors and genuinely wondered if my stubbornness was just sunk-cost fallacy with a diploma attached.

Here's where I landed: the fundamentals haven't changed. Data structures, algorithms, systems design, networking, concurrency — these are the load-bearing skills underneath every AI tool, every cloud service, every piece of software that matters. AI didn't make these skills obsolete. It made them more valuable, because now you need them to evaluate what AI produces.

What I'm doing differently:

Building AI-adjacent skills on top of my CS core. Not switching to an AI major — learning to use AI tools within the context of systems I already understand. There's a difference between "I understand transformers" and "I can build a reliable system that uses a transformer." The second one pays.

Contributing to open source while the competition thins out. With enrollment dropping and fewer students contributing, the barrier to making a meaningful contribution is lower than it's been in years. One merged PR during a talent drought is worth five during a gold rush.

Treating the pessimism as a feature, not a bug. When 70% of your peers are pessimistic, a lot of them are applying with less confidence, less energy, and less conviction. Showing up with genuine enthusiasm and a clear thesis about why you chose this field is an edge you can't manufacture.

A note on the "AI degree" question

If you're considering switching from CS to a new AI-specific program, ask yourself: what does this degree teach that I can't learn by taking AI/ML electives within my existing CS program? A CS degree gives you operating systems, networking, databases, and software engineering principles — the foundations that make AI skills useful in production. An AI degree without those foundations is like learning to drive without understanding what an engine does. You might be fine on a clear road. But the first time something breaks, you're stuck.

The bottom line

CS enrollment is falling because students are reading headlines instead of data. The headlines say "AI will replace developers." The data says starting salaries are up 7%, AI companies are hiring juniors for the first time, and the companies that tried hardest to replace humans with AI are quietly hiring humans again.

I'm staying in CS. Not because I'm ignoring the risks — I've written about them honestly. But because the students leaving right now are creating exactly the kind of supply-demand imbalance that makes the ones who stay disproportionately valuable.

If you're a CS student thinking about switching: finish the degree. Take the AI electives. Build things that work in production, not just in a Jupyter notebook. The trough is temporary. The foundations are permanent. And in two years, when the headlines flip to "CS graduates in high demand as AI boom creates engineering shortage," you'll be glad you didn't flinch.