For the last fifteen years, the safe answer to “what should my kid major in?” was computer science. Strong job market. High starting salaries. Translatable to almost any industry. Parents pushed students toward it. Students applied in record numbers. By the Class of 2026, eleven percent of all college students were majoring in CS, up from seven percent a decade earlier.

That run is over. And the families who are paying attention are now making the next mistake.

What just happened to CS

Three things broke at once.

First, the entry-level job market for CS graduates collapsed. According to Handshake’s analysis of Class of 2026 hiring, software engineering roles fell from a top five most-posted entry-level job every year between 2018 and 2023 to ninth place in the 2024-25 cycle. Tech companies cut more than 260,000 jobs in 2023 alone. New CS graduates are taking six-plus months to land a first role.

Second, generative AI came for exactly the work that junior software engineers used to do. Boilerplate code, basic debugging, documentation, the stuff a new graduate gets paid eighty thousand dollars to learn on the job, is now done faster and more accurately by an AI model. Seventy percent of Class of 2026 CS majors describe themselves as pessimistic about their careers. Among those, two-thirds explicitly cite AI as a reason. This is the demographic actually doing the work, telling you the work is changing.

Third, students started voting with their feet. The University of California system, a leading indicator for the rest of the country, reported a six percent drop in CS enrollment in the 2025-26 cycle, on top of a three percent decline the year before. It is the first sustained decline since the dot-com crash. The Computing Research Association’s most recent survey found sixty-two percent of universities reporting falling CS enrollment.

The decade-long surge has reversed. That is not a small story. And it has been on the front page of TechCrunch and the Washington Post in the last few weeks.

Where the money is moving

Here is where it gets interesting for a junior parent right now. Students are not leaving technology. They are moving sideways into majors that look more durable.

USC, Columbia, Pace University, New Mexico State, and the University of South Florida all launched dedicated AI majors this year. MIT’s “AI and decision-making” track is now its second-largest major. Buffalo opened seven new AI-focused undergraduate programs and got two hundred applications before the doors opened. Cybersecurity programs are absorbing thousands of students. Data science is everywhere. Mechanical and electrical engineering, the parts of engineering that build things AI cannot yet replace, are growing.

This migration is rational. A student in 2026 looking at a CS program that is mostly about writing code, when AI now writes that code competently, is right to ask whether the major is still the right vehicle. So the strategic move looks obvious. Skip CS. Apply to AI, or data science, or cybersecurity. Same career outcome, more durable framing.

That is where the trap is.

The trap

Every wave of “the new safe major” creates the next bubble. CS in 2010 was a smart pick because few students were doing it and the demand was structural. CS in 2024 was the most-applied-to major at every selective school in America and admit rates at top programs cratered to three to six percent. The major did not change. What changed was that everyone showed up at once.

AI is now in the position CS was in around 2015. Demand looks structural. The story is plausible. Programs are launching faster than they can be staffed. And every parent and student who reads the same TechCrunch article you read this morning is reaching the same conclusion: switch from CS to AI.

By the 2027 application cycle, AI will be the most over-applied intended major at any school that offers it. By 2028, admit rates to AI programs at selective schools will look like CS admit rates today, three to six percent, with applicants who all have research experience, hackathon wins, and personal projects deployed on GitHub. The strategic move that made sense for the Class of 2018 will be the wrong move for the Class of 2030.

This is not a prediction that requires unusual foresight. It is the same dynamic that produced the CS bubble in the first place, running one cycle later, on a different bucket.

What this means for a junior right now

Three things to think about, if you are applying in the 2027 cycle.

One. The intended major box on the Common App is a strategy question, not a paperwork question. Selective schools that admit by program, including Cornell, Carnegie Mellon, Berkeley, and Michigan, have admit rate gaps of two to three times across colleges within the same university. Cornell’s College of Arts and Sciences admits at nearly double the rate of its College of Engineering, and that gap has widened in 2026, not narrowed. Berkeley’s L&S Computer Science path admits at thirteen to fifteen percent. Berkeley’s EECS program admits under five percent. Same university. Same CS degree. Different application odds.

Two. Applying to CS now might actually be the contrarian move. The pool is thinning. The seats are not shrinking yet. A junior who genuinely loves the discipline, not the salary, the discipline, has better admit odds in 2027 than the previous five years. That window will close. It is open right now.

Three. If your student is genuinely AI-curious, the strategic move is not to apply as an AI major. It is to apply as math, applied math, statistics, cognitive science, or, at schools without an AI track, CS with research experience that signals AI-relevant work. These paths route to the same graduate school options, the same job market, and dramatically better admit rates than whatever the hot bucket happens to be in 2027.

The actual question

The question worth asking your junior is not “what is the safe major now?” The question is “what does this student actually want to be doing at twenty-six years old, regardless of what the application strategy is?” Because every other variable, the major, the school, the application angle, flows from that. Every other approach is just chasing the last bubble’s headline one cycle late.

If you are rethinking your junior’s college list and want to talk through what their actual options look like given their interests and the current cycle data, that is a conversation worth having early. The decisions that matter most happen between May and September of junior year. We are in May.