AI Impact on Computer Science Enrollment Hits a Historic Low in 2025

For the first time since the dot-com crash, American universities are watching the numbers flip on AI impact computer science enrollment. CS majors fell 8.1% in fall 2025, the sharpest single-year drop in any field since at least 2020. Meanwhile, AI-specific degree programs are filling classrooms faster than schools can hire professors. We have also seen this shift coming for months across the industry, and the fresh data now confirms a deep rebalancing in how young people choose to study technology and plan their careers.

The enrollment numbers behind the story

The scale of the drop is striking. According to the Computing Research Association, 62% of computing programs nationwide saw undergraduate enrollment fall this year. About 54,000 fewer CS majors are studying today than last year. As a result, computer science slipped from the fourth largest major to sixth, behind business, health, and liberal arts. The total college population, however, grew 2%, so this is not a general slowdown in higher education. In fact, students appear to be actively choosing alternatives rather than skipping college altogether.

The University of California system lost 6% of its CS majors in fall 2025, on top of a 3% decline in 2024. That pattern is not a single-year blip. It shows two straight years of contraction at one of the most competitive university systems in the country. In contrast, UC San Diego saw gains after launching a dedicated AI major. That single exception points to the main story: students still want to study technology, but many no longer see traditional CS as the smartest path.

Why are students voting with their applications?

Students are steering away from traditional CS because they see AI tools writing code, senior engineers producing less code themselves, and entry-level tech hiring falling fast. In short, the numbers stopped adding up for a four-year CS degree.

Students are making careful calculations, not panicking. They have watched senior engineers at Spotify say they have not written a line of code since December 2025. They have read that Google leadership pegs AI-generated code at roughly half of all new output, with the real share "much, much higher." Anthropic reportedly writes 70% to 90% of its code with AI help. Microsoft pegs its share at 30%. For an 18-year-old choosing a major, those headlines carry weight.

The job data makes the picture worse. Unemployment for recent CS graduates sits at 6.1%, well above the 4.8% average for all new graduates. Entry-level tech hiring fell 25% year over year in 2024. Worse still, employment for software developers aged 22 to 25 has dropped nearly 20% since late 2022. These are not small numbers, and students see them clearly. Parents and high school counselors are also passing the same stats along during college planning meetings.

The automation versus augmentation split

A sharp pattern shows up in the data. Stanford's Digital Economy Lab found that young workers lose jobs fastest in roles where AI automates work, not where it augments it. Junior developers historically wrote boilerplate code, patched known bugs, and built features from a spec. Those are the exact tasks modern coding agents handle best. Senior roles, by contrast, require judgment, architecture choices, and messy real-world calls, and those jobs are still growing.

Hiring managers have shifted their stance too. Surveys show 70% think AI can do intern work, and 57% trust AI output over work from a new grad. Even more bluntly, 37% say they would rather use AI than hire a recent graduate. Those views shape budgets, and budgets shape openings. For new graduates, the first step into a software career now requires stronger signals of readiness than it did a few years ago.

AI majors are absorbing the demand

The migration is clear. Fresh AI degree programs are sprouting across American universities and pulling applicants away from traditional CS. Notable 2025 and 2026 launches include:

• MIT: the AI and Decision-Making major is now the second-largest program on campus.
• University of South Florida: a new AI and cybersecurity college welcomed 3,000 students in its first fall.
• University at Buffalo: the AI and Society department saw more than 200 applications before opening, with seven undergraduate tracks.
• USC, Columbia, Pace, and New Mexico State: all announced new AI programs for 2026.

Similar growth is showing up in adjacent fields. Data science and data analytics programs enrolled more than 35,000 students in fall 2025, up from only a few hundred in 2020. Engineering overall rose 7.3%, with mechanical up 11% and electrical up 14%. In short, students still want technology careers; they simply want titles that feel less exposed to coding agents. Vocational tracks in engineering technologies, mechanic and repair work, and health professions are also climbing, with growth between 8% and 11%.

How universities are rewriting the CS curriculum

Meanwhile, schools are not standing still. The University of Washington's Paul G. Allen School announced that coding, or the translation of a precise design into software instructions, is dead because AI now handles that layer. Its leaders argue that Allen has never graduated coders and always produced engineers. That framing now drives curriculum redesign across the country. For example, courses are moving toward system design, security, product judgment, and managing AI tools, rather than manual typing of syntax. Assignments increasingly ask students to validate and debug AI output rather than produce code from scratch.

Institutional adoption is moving fast as well. According to the 2026 AI Index from Stanford HAI, institutional AI adoption in higher education jumped from 49% in 2024 to 66% in 2025. Forty-three percent of schools now include AI in their strategic plans. K-12 is lagging behind, though. Only half of middle and high schools have AI policies, and just 6% of teachers call those policies clear. That gap will need to close quickly if younger students are going to arrive on campus ready for modern CS.

What this means for the future tech workforce

This enrollment shift creates two realities at once. The short-term win is cheaper output. The long-term risk is a missing middle. Specifically, if entry-level hiring stays frozen, the pipeline feeding senior talent will thin out. In a few years, companies may struggle to find the mid-level engineers they now take for granted. As a result, we have seen smart firms respond by investing in structured AI talent and upskilling programs, pairing new hires with coding agents from day one, and redefining the junior role around supervision and review rather than line-by-line writing.

For business owners thinking about workflow automation, the message is practical. Tooling is ready, models are cheaper every quarter, and the talent pool knows AI from day one. Firms that pair strong senior judgment with AI-powered teams will run leaner and move faster than those still waiting for traditional hires to fill out an org chart. Teams ready to explore where AI fits can book a free consultation to scope a practical starting point. The students watching from the sidelines have already made their choice; the rest of the market is catching up quickly.