Three years ago, companies were fighting over engineers. Signing bonuses, remote offers from day one, recruiters sliding into your DMs before you even updated your LinkedIn. It was an employee's market, and if you could write a for loop, someone wanted to pay you six figures to do it. Now the power has flipped. It's an employer's market, and the shift happened fast enough to give you whiplash.

What changed

The short answer: supply exploded, demand contracted, and AI gave companies a convenient reason to accelerate both trends at once. The longer answer starts with COVID. When the pandemic hit, tech became the safe bet. Remote work was suddenly the default. Tech salaries looked untouchable. Every parent, career counselor, and LinkedIn influencer pointed in the same direction: study computer science. And people listened. CS enrollment surged worldwide. Universities couldn't build lecture halls fast enough. Bootcamps multiplied. The message was clear: learn to code, and you'll never struggle to find a job. Then the graduates arrived, right as the music stopped.

The supply problem

The pipeline that COVID supercharged didn't have an off switch. Students who enrolled in 2020 and 2021, riding the wave of a booming tech market, are graduating now into a completely different reality. The numbers paint a stark picture. The unemployment rate for recent computer science graduates sits at 7.0%, well above the 4.2% average for all early-career graduates, according to the Federal Reserve Bank of New York. That puts CS graduates in the same unemployment bracket as performing arts majors, a comparison that would have been unthinkable five years ago. The underemployment rate is even more telling. As of Q4 2025, 42.5% of recent graduates are working in jobs that don't require a college degree. Nearly half of the people who spent four years studying algorithms and data structures are doing work that doesn't use any of it. Meanwhile, TechCrunch reported on what they called "the great computer science exodus," with students now actively pivoting away from CS and toward AI-specific programs or other fields entirely. The Atlantic called it the bursting of the computer science bubble. Enrollment is cyclical, and we're watching the downswing in real time.

The demand problem

On the other side of the equation, hiring has collapsed, especially for entry-level roles. Ravio's 2025-2026 tech job market reports found that junior-level hiring fell 73.4% year-over-year. That's not a dip. That's a crater. Tech job listings overall are down 35% from pre-pandemic levels, according to Indeed, and there are no signs of a rebound. The market is in what analysts are calling a "holding pattern," stable at low levels with no boom-bust cycle on the horizon. A Stanford Digital Economy Lab study found that early-career software developers aged 22 to 25 have seen roughly a 20% decline in employment in AI-exposed roles, even after controlling for interest rates and firm-specific shocks like COVID over-hiring. This isn't a temporary correction. It's structural. Companies aren't just hiring less. They're hiring differently. The demand has shifted hard toward specialists, particularly in AI and machine learning. Ravio's data shows AI/ML hiring grew 88% year-on-year in 2025, while general software engineering roles stagnated or shrank. If you're a generalist fresh out of school, you're competing for a shrinking pool of positions against people with years of experience who were laid off from the last round of cuts.

The AI excuse

Here's where it gets uncomfortable. AI is genuinely changing what companies need, but it's also become the most convenient excuse in corporate history for cutting headcount. A Harvard Business Review survey of over 1,000 global executives found that AI-related layoffs are happening almost entirely "in anticipation of AI's impact," not because AI is actually performing the work yet. Companies are cutting based on potential, not performance. A separate study found that 60% of companies emphasize AI when announcing layoffs, even when budget constraints and restructuring are the primary drivers. The BBC reported on the pattern bluntly: tech CEOs suddenly love blaming AI for mass job cuts. Google, Amazon, Meta, Pinterest, Atlassian, they've all pointed to AI as the reason for shrinking their workforce. The stock market rewards it every time. Announce layoffs, invoke AI, watch the share price recover. But the reality on the ground is messier. Most companies aren't replacing developers with AI. They're using AI as cover for cost-cutting that was going to happen anyway, driven by rising interest rates, slower growth, and the end of the zero-interest-rate era that inflated tech valuations for a decade.

The race to the bottom

When supply outstrips demand this dramatically, the economics get ugly fast. If two candidates have equivalent skills, companies take the cheaper one. That's always been true, but it used to be a tiebreaker. Now it's the primary filter. When you have hundreds of applicants for every junior role, price becomes the differentiator. It gets worse. If one candidate is less experienced but can use AI tools effectively and costs 40% less, companies are increasingly choosing that combination. Why pay for a mid-level developer when you can hire someone cheaper and give them Copilot? The logic is seductive, even if the long-term consequences are severe. This creates a domino effect. Wages compress. Experienced developers take roles beneath their skill level to stay employed, which pushes entry-level candidates further down or out entirely. The people who do get hired accept lower pay because the alternative is no pay. More work gets piled on fewer people because the headcount budget is frozen but the roadmap isn't. Dice's research found that 80% of tech professionals have applied to a "ghost job" in the past year, a posting that was never intended to be filled. Over half have applied to jobs beneath their skill level just to stay employed. Confidence in tech's long-term growth has dropped to 60%, down from 80% in previous years.

The joke that isn't funny

There's a meme that keeps circulating: the funniest prank is convincing someone to study for four years, only for AI to replace them the day they graduate. It's funny until you're the punchline. The irony cuts deep because there's truth in it. Students who enrolled in CS programs during the COVID boom were responding to rational market signals. Tech was booming. Salaries were high. Demand seemed infinite. By the time they graduated, the market had inverted. The skills they spent four years acquiring are now competing with tools that can generate code in seconds. But here's the thing: the joke is also misleading. CS graduates are still projected to earn starting salaries of $81,535, up nearly 7% from last year according to NACE's 2026 salary survey. The BLS projects software developer employment to grow 15% from 2024 to 2034, much faster than average. The field isn't dying. It's restructuring. The problem isn't that computer science is worthless. The problem is that the market can't absorb the volume of graduates at the pace they're arriving, especially when companies are simultaneously cutting entry-level positions and raising the bar for who they'll hire.

What's actually happening

The job market isn't broken in one direction. Several forces are colliding at once. COVID created a supply glut by making CS the obvious career bet. The end of cheap money forced companies to cut costs and hire more cautiously. AI gave companies both a real reason and a convenient excuse to restructure. And the shift toward specialist roles means the generalist pipeline that universities built is mismatched with what employers want. The result is a market where there are simultaneously too many candidates and not enough of the right ones. Companies complain about talent shortages while graduates can't get callbacks. Both things are true, and neither side is wrong. Workers with genuine AI skills earn a 56% wage premium over colleagues in the same role without those skills, according to PwC's 2025 Global AI Jobs Barometer. AI/ML roles command a 12% salary premium even at the individual contributor level. The market isn't shrinking uniformly. It's bifurcating. If you're on the right side of the split, opportunities are abundant. If you're on the wrong side, it feels like the industry abandoned you.

The longer game

This has happened before. CS enrollment is cyclical. It crashed after the dot-com bust, recovered, crashed again in 2008, and recovered again. Each time, the recovery brought enrollment higher than the previous peak. There's no reason to believe this cycle is different in kind, even if AI makes it different in degree. But the transition period is brutal, and pretending otherwise doesn't help anyone. The students who started CS in 2020 expecting a smooth ride into a six-figure job deserve honesty about what they're walking into. The experienced developers who got laid off deserve better than being told AI made them redundant when the real reason was a cost cut dressed in a press release. The job market isn't dead. But the version of it that existed in 2021, where demand was infinite and supply was scarce, that's gone. What's replacing it is harder, more competitive, and less forgiving of people who can't differentiate themselves. The question isn't whether there will be software jobs. There will be, lots of them. The question is whether the path to those jobs will remain accessible, or whether we're building a market that only works for people who were already ahead.

References