↓ Skip to main content
Miles Wallace

Computer Science Enrollment Decline & AI Migration

·1968 words
Miles Wallace
Author
Miles Wallace

CS Enrollment Decline & the AI Migration

For the first time since the dot-com crash of the early 2000s, computer science enrollment at U.S. universities is in measurable decline. Across the University of California system, one of the most influential public university networks in the country. CS majors dropped 6% in 2025 to 12,652 students, following a 3% decline in 2024. That figure still represents nearly double the number of CS majors from a decade ago, making the reversal all the more striking. Harvey Mudd College, a school Where CS and STEM-related fields make up more than half of its graduates. Reported a 27% drop in applications from students listing computer science as their first-choice major over the last four years. A survey by the Computing Research Association found that 62% of computing departments nationally reported declining undergraduate enrollment in fall 2025. The granular breakdown is telling: 40% of those units reported slight decreases of 0–10%, 29% reported moderate decreases of 11–20% and 31% reported enrollment declines of more than 20%. Among declining departments, the average drop was 11–15%. At the graduate level, the contraction is even sharper. CS graduate enrollment fell 14.0% nationally, compared to -3.6% at the undergraduate level. These are not rounding errors or statistical noise, they represent a structural shift in how students and families perceive the value of a CS education in the age of artificial intelligence. The decline is happening against a backdrop of rising overall college attendance: total U.S. postsecondary enrollment climbed 2% in fall 2025, reaching 19.4 million students. CS is contracting even as the pool of students grows.

The Job Market Has Fundamentally Changed
#

The traditional pitch for a CS degree was straightforward: go to college, learn to code, get a well-paying job at a tech company with excellent benefits and strong job security. That pitch no longer holds. Following waves of mass layoffs in 2022 and 2023, the technology sector shed over 150,000 jobs in 2024, followed by an additional 122,549 layoffs across 257 companies in 2025. The Wall Street Journal described the situation bluntly, there is “no more red carpet” for CS graduates. The lavish perks that once defined Silicon Valley recruiting culture, free meals, on-site amenities, premium signing bonuses have largely vanished. Entry-level software engineering roles, the traditional first landing spot for CS graduates, have been among the hardest hit. Entry-level hiring at the 15 biggest tech firms fell 25% from 2023 to 2024. Overall programmer employment fell 27.5% between 2023 and 2025. The unemployment rate for recent CS graduates reached 6.1% in 2025, nearly double the rate of philosophy majors. A striking inversion of the expected credential hierarchy. Generative AI tools like GitHub Copilot and ChatGPT can now produce functional code for routine tasks faster and cheaper than a junior developer, making those entry-level positions economically redundant from many employers’ perspectives.

AI Is Displacing the Work, Not Just the Workers
#

The impact of AI on software development is not simply that companies are hiring fewer people, it is that the nature of the work itself is being restructured. A senior engineer who previously needed a team of 15 to build and maintain a business application can now accomplish the same with a team of five, using modern development tools, open-source platforms and AI-assisted code generation. Entry-level programming — writing CRUD operations, pulling records from databases, implementing basic business logic, is precisely the kind of work that large language models handle well. The result is a hollowing out of the bottom of the software engineering career ladder. Junior roles that once served as entry points into the industry and training grounds for future senior engineers are disappearing. Creating a structural gap between the CS graduate and the kinds of roles that still require deep human expertise. Notably, demand for engineers skilled specifically in AI/ML and infrastructure is rising. The displacement is concentrated at the generalist entry-level layer, not across all of software engineering. The BLS still projects the software development sector to grow 17% by 2033; the question is what shape of workforce fills that growth.

The Parental Factor and the ROI Calculation
#

Parents have played a significant role in driving CS enrollment over the past two decades. Watching figures like Bill Gates, Jeff Bezos, Mark Zuckerberg, Steve Jobs, Larry Page and Sergey Brin generate extraordinary wealth from software. Many families concluded that computer science was the modern equivalent of medicine or law, a reliable, high-income professional path. That consensus is now cracking. Admissions consultants and university officials report that parents who once aggressively pushed their children toward CS are now reflexively steering them toward adjacent fields mechanical engineering, electrical engineering or other technical disciplines perceived as harder to automate. The concern is not irrational. Families are watching headlines about tech layoffs and AI-driven job displacement and recalculating the return on investment of a four-year CS degree that can cost anywhere from $50,000 to over $300,000 in total. When the expected payoff is uncertain, the risk calculus changes.

The Academic and Personal Demands of CS Have Not Diminished
#

Even setting aside job market concerns, computer science has always been a demanding and unforgiving field of study. It requires a specific cognitive temperament, a tolerance for abstraction, an affinity for logical problem-solving and the psychological resilience to spend days or weeks stuck on a single problem without giving up. For students who do not have a genuine passion for the material, the coursework can be demoralizing. Programming assignments in academic settings often function like extremely difficult logic puzzles with strict deadlines and the frustration of debugging a complex algorithm for hours without resolution is not something everyone can endure. With the job market offering less certain rewards than before, more students are choosing not to endure it. Particularly those who were motivated primarily by salary expectations rather than intellectual curiosity about computing itself.

Not a Retreat From Tech — A Migration Toward AI
#

The critical nuance in this enrollment story is that students are not abandoning technology as a field. They are migrating toward AI-specific programs that feel more directly relevant to where the industry is heading. UC San Diego, the only UC campus that bucked the downward trend, did so precisely because it launched a dedicated AI major in fall 2025. MIT’s “AI and Decision-Making” track is now the second-largest major on campus. The University of South Florida enrolled more than 3,000 students in a newly created AI and cybersecurity college in its first semester. The University at Buffalo launched an “AI and Society” department offering seven specialized undergraduate degree programs and received over 200 applicants before it had even fully opened. Columbia University, USC, Pace University and New Mexico State University are all launching AI-focused degrees in 2026. The pattern is clear: where traditional CS enrollment falls, AI program enrollment rises to fill the void.

The China Comparison and U.S. Universities’ Slow Response
#

The United States’ rocky transition stands in sharp contrast to how China has approached AI in higher education. As MIT Technology Review reported, Chinese universities have moved rapidly to treat AI literacy as foundational infrastructure rather than an elective competency. Nearly 60% of Chinese students and faculty now use AI tools multiple times daily. Zhejiang University has made AI coursework mandatory across disciplines. Tsinghua University created entirely new interdisciplinary AI colleges. The philosophical framing is different: in China, AI is not a threat to be managed but a capability to be mastered. In the U.S., universities are scrambling to catch up, but the transition is uneven. UNC Chapel Hill Chancellor Lee Roberts, a former finance executive, described the faculty landscape as a spectrum. Some leaning forward into AI integration, others with “their heads in the sand.” When UNC announced a merger of two schools to create an AI-focused entity, it drew significant faculty pushback. The institutional inertia of traditional academia is slowing a transition that market forces are demanding.

When a CS Degree Still Makes Sense
#

Despite the headwinds, a computer science degree remains a powerful credential in specific contexts. For students with a genuine intellectual passion for computing, those who enjoy the challenge of hard problems as a matter of curiosity rather than obligation. The degree still opens doors that few other programs can match. Students aiming for careers in AI and machine learning research will need the deep mathematical and algorithmic foundation that CS programs provide; someone still needs to design, train, evaluate and improve the models that everyone else is using. Robotics and autonomous systems represent another strong path, a field that draws on computer science, electrical engineering and mechanical engineering simultaneously. Also, one likely to grow significantly as demographic decline forces labor-intensive industries to automate. For students planning to attend graduate school, law school or move into product and research roles at technology companies, an undergraduate CS degree provides a rigorous, highly transferable analytical foundation.

The Skills Gap That CS Programs Often Ignore
#

One underappreciated dimension of this conversation is that traditional CS programs have long underinvested in communication and writing skills. When experienced engineers and hiring managers evaluate candidates, the ability to write clearly, reason about ambiguous problems and communicate technical ideas to non-technical stakeholders is consistently identified as a differentiating factor. A philosophy major who can code is often more valuable than a CS major who cannot write. This is not a knock on CS curricula, it is a reflection of how narrow specialization creates blind spots. Students who recognize this gap have an opportunity: double-majoring or minoring in a liberal art alongside CS or vice versa. Produces a candidate profile that is genuinely rare and genuinely useful. The combination of technical competency and humanistic reasoning is precisely what the AI era demands, since human judgment, ethical reasoning and cross-domain thinking are among the last things that generative AI struggles to replicate reliably.

The Bigger Picture
#

What we are witnessing is not a generation turning its back on technology. It is a generation recalibrating in real time, trying to identify where human expertise will remain valuable as AI reshapes the economic landscape of software development. The students choosing AI programs over traditional CS programs are not making a worse bet — they may be making a better one, if those programs genuinely prepare them to work at the frontier of the technology rather than in roles that the technology itself is replacing. The universities that adapt fastest — by integrating AI literacy across all disciplines, launching interdisciplinary AI programs and rethinking what a CS education should produce. Will define what technical education looks like for the next generation. Those that resist, clinging to curricula designed for a world where entry-level coding jobs were abundant and AI was a research curiosity. Risk producing graduates who are technically credentialed but professionally misaligned with the market they are entering.

Sources
#