By Thorsten Meyer | October 29, 2026 | ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
For years, “learn to code” was the safest career advice a parent could give. This fall, students are walking away from it. Early data released today by the National Student Clearinghouse Research Center show enrollment in computer and information science programs down 12.4% at four-year colleges, 12.3% at two-year colleges and 10.4% in graduate programs, compared with last fall. None of the other large fields in the report fell that hard.
The coverage already has its villain. Bloomberg’s story on the data says AI has made it harder for computer science graduates to find jobs “as companies eliminate many entry-level coding roles.” That is part of the picture. But a falling number and a popular explanation are not the same thing as a cause. Visa rules, the hangover from the 2021-2022 hiring boom, remote work and a shift toward newer programs all point in the same direction at the same time.
So here is what the enrollment data actually show, what the job data show, which explanations hold up, and what it means for companies and for students, parents and career switchers.
Key takeaways
- The drop: preliminary fall 2026 data show computer and information science enrollment down 12.4% at four-year colleges (undergraduate), 12.3% at two-year colleges and 10.4% at the graduate level. Over two years, the declines are 18.4%, 17.5% and 27.1%.
- Not a flight from STEM: at four-year colleges, engineering grew 6.1% and health professions 6.2% this fall. Students are leaving computer science, not technical study.
- The graduate drop is largely an international story: at research universities, about half of computer science master’s students are nonresidents, and 86% of computing departments in a spring 2026 survey said visa delays and denials hurt their incoming graduate class.
- The entry-level job market is genuinely weak: Indeed’s software development postings index stands at about 81 (February 2020 = 100), and only 4.5% of software development postings in early 2026 were entry-level. Recent computer science graduates had a 7.0% unemployment rate in the latest by-major data.
- AI’s role is real but not proven as the main cause: a Stanford study finds employment of 22- to 25-year-olds in AI-exposed jobs 19% below where it would otherwise be. A New York Fed analysis attributes most of the rise in young graduates’ unemployment to remote work, and its timing predates generative AI.
- For companies: record graduating classes now, smaller ones from about 2029. Redesign junior roles instead of deleting them.
- For students and switchers: computer science still pays well above average. Build skills that combine code with judgment, verification and a domain.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
Apple iPad Air (4th Generation, 10.9-inch, Wi-Fi, 256GB) – Space Gray
As an affiliate, we earn on qualifying purchases.
What the Clearinghouse actually measured
The Preliminary Fall Enrollment Trends report, published on October 29, 2026, is an early look. It is based on 7.92 million enrollments reported by 46.0% of participating institutions as of September 24, 2026. The Clearinghouse compares the same panel of schools across three years and reports percentage changes, not totals. The final numbers come in January 2027 and can differ.
The overall picture is calm. Total enrollment is up 0.2%. Undergraduate enrollment is up 0.6%, after 2.4% last fall. Graduate enrollment is down 1.6%, mainly because master’s enrollment fell 3.7%. “That stability doesn’t tell the whole story,” said Matthew Holsapple, the Clearinghouse’s senior director of research, in the release.
Computer science is where the story is. The field (“Computer and Information Sciences and Support Services” in the federal program classification) fell at every level:
- Undergraduate, four-year colleges: down 6.9% last fall, down 12.4% this fall, down 18.4% over two years.
- Undergraduate, two-year colleges: down 5.9%, then 12.3%, 17.5% over two years.
- Graduate: down 18.6%, then 10.4%, 27.1% over two years.
The undergraduate decline is accelerating. The graduate decline is slowing, after a very steep first year. Keep that difference in mind. It is the first clue that more than one force is at work.
Infographic 01 · The turn
Computer science is down at every level, while engineering and health grow
Early fall 2026 data from 46% of U.S. institutions. Percentage changes for the same panel of schools; final figures follow in January 2027.
Every level is down
Change in enrollment versus the prior fall. The undergraduate drop is speeding up, the graduate drop is slowing after a steep first year.
Undergraduate, 4-year
Undergraduate, 2-year
Graduate
Same 4-year colleges, other fields
Undergraduate enrollment change, fall 2026 vs. fall 2025. Students are leaving one field, not technical study.
- Computer & information science−12.4%
- Liberal arts & humanities−4.5%
- Visual & performing arts−4.2%
- Business+0.1%
- Biology+3.8%
- Engineering+6.1%
- Health professions+6.2%
Source: National Student Clearinghouse, Preliminary Fall Enrollment Trends, 29 Oct 2026 (dashboard Figure 3.5; prior-year values from the same panel). Unweighted, preliminary, subject to revision. Field names shortened. For scale: undergraduate computer science at 4-year colleges grew from about 499,000 (2020) to about 660,000 (2024), then fell 8.1% to about 606,000 in fall 2025 (NSC Final Fall Enrollment Trends 2025).
Powered by Thorsten Meyer AI
At the same four-year colleges, engineering grew 6.1% this fall and 14.5% over two years. Health professions grew 6.2%. Business was flat. Students are still choosing technical and career-focused fields. They are choosing them instead of computer science.
Zoom out: the boom before the drop
The drop follows a boom. In the Clearinghouse’s final fall 2025 data, undergraduate computer science enrollment at four-year colleges grew from about 499,000 in fall 2020 (520,000 in 2021, 574,000 in 2022, 629,000 in 2023) to about 660,000 in fall 2024, then fell 8.1% to about 606,000 in fall 2025. Graduate enrollment in the field fell 14.0% in fall 2025, to about 190,000.
Rough arithmetic of my own: if this fall’s preliminary 12.4% held for the final count, undergraduate computer science at four-year colleges would land near 530,000, close to its 2021 level. That mixes a preliminary rate with final counts, so treat it as an order of magnitude, not a forecast.
The pipeline data from the Computing Research Association’s 2025 Taulbee Survey, which covers PhD-granting computing departments in the U.S. and Canada, show the same turn. New computer science majors fell 13% to 33,490. At the same time, these departments awarded a record 41,858 bachelor’s degrees, because degrees reflect students who enrolled four years earlier. The drop has not reached the job market yet. It will.
What students see in the job market
Students are not reacting to nothing. Three data sets describe the market they are looking at.
Job postings. Indeed’s software development postings index (February 1, 2020 = 100, seasonally adjusted) peaked at about 234 at the end of February 2022. By early 2023 it was at about 130, by early 2024 about 73, and it bottomed at about 61 in May 2025. From peak to low, it lost about three quarters. On October 23, 2026, it stood at about 81: up roughly a quarter from a year earlier, but still about a fifth below its pre-pandemic level and about two-thirds below the peak.
Who those postings are for. Indeed’s Hiring Lab found in July 2026 that only 4.5% of software development postings in the first quarter of 2026 were entry-level, the lowest share of any sector in its comparison. Across all jobs, senior-level postings rose almost 15% between May 2025 and May 2026, while entry-level postings were down 6.2% from January 2025.
Outcomes for graduates. In the New York Fed’s labor market data for recent graduates (ages 22 to 27; by-major figures for 2024), computer science majors had a 7.0% unemployment rate and computer engineering majors 7.8%, against 4.2% for all majors. In a fall 2025 survey of computing departments, 66% agreed that their 2025 graduates were having difficulty finding a job. One department head in a spring 2026 survey reported that 90% of their 2024 master’s graduates had a job in the field within six weeks. For the class of 2025, it took eight months to reach 90%.
Infographic 02 · What students see
Most of the collapse in software job ads came before AI coding tools
The postings index more than doubled into 2022, crashed through 2024 and is recovering slowly. Entry-level roles remain scarce.
Software development job postings on Indeed
Index, 1 February 2020 = 100, seasonally adjusted, first reading of each month plus the latest value. Postings are job ads, not hires.
Sources: Indeed Hiring Lab job postings data, CC BY 4.0, through 23 Oct 2026; Indeed Hiring Lab, 23 Jul 2026; New York Fed, recent-graduate data (by major: 2024); Stanford Digital Economy Lab, Aug 2026. Monthly values are the first daily reading of each month; peak and low from the daily series.
Powered by Thorsten Meyer AI
Five suspects: correlation versus causation
When two lines move together, the question is which one pushes. Here are the five explanations on the table, and what the evidence says about each.
Infographic 03 · Five suspects
Why are students leaving computer science? Five explanations, checked
Several forces push in the same direction at the same time. Here is what the evidence supports for each.
- 1Strong evidence (timing)
Hiring hangover after 2022
Indeed's software postings index fell from about 234 (Feb 2022) to about 73 (Jan 2024), mostly during the rate hikes and post-pandemic cutbacks, before AI coding tools spread.
- 2Strong for graduate programs
Visas and international students
About half of CS master's students at research universities are nonresidents. 86% of computing departments say visa delays and denials hurt their new graduate class. New international enrollment fell 17% in fall 2025.
- 3Real, not proven as main cause
AI doing junior work
Stanford: 22- to 25-year-olds in AI-exposed jobs 19% below trend, mostly through less hiring. But the authors call it descriptive, and a New York Fed analysis attributes 64% of young graduates' rising unemployment to remote work.
- 4Partial, size unknown
Moving to neighbouring programs
Departments report growth in computer engineering, data science, cybersecurity and AI. Engineering at 4-year colleges is up 6.1%. Part of the move leaves the 'computer science' statistics.
- 5Weak for CS specifically
Cost and demographics
Affordability pushes students toward community colleges. But cost affects every major and cannot explain why CS fell while engineering and health grew.
Verdict: layered causes. AI is a real and growing factor, not yet the proven main cause.Thorsten Meyer AI reading of the evidence below. Ratings are editorial judgments, not statistical tests.
Sources: Indeed Hiring Lab job postings data, CC BY 4.0, through 23 Oct 2026; CRA Taulbee Survey 2025; CRA CERP undergraduate survey, Oct 2025; CRA CERP graduate survey, Apr 2026; IIE Fall 2025 Snapshot; Stanford Digital Economy Lab, Aug 2026; New York Fed Liberty Street Economics, 1 Jun 2026; National Student Clearinghouse, Preliminary Fall Enrollment Trends, 29 Oct 2026.
Powered by Thorsten Meyer AI
1. The hiring hangover after 2022
Look at the timing of the postings collapse. Most of it happened between early 2022 and early 2024, as interest rates rose and tech companies cut back after pandemic-era over-hiring. ChatGPT launched at the end of November 2022, and capable coding agents arrived much later. Indeed itself lists the Fed’s rate hikes and post-pandemic “right-sizing” among the competing explanations. My reading: the first and biggest blow to junior software hiring was not AI. Students who started college in 2025 and 2026 grew up watching that slump.
2. AI doing junior work
This is the explanation in the headlines, and there is real evidence for it. The August 2026 revision of Stanford’s “Canaries in the Coal Mine?” paper by Erik Brynjolfsson, Bharat Chandar and Ruyu Chen uses ADP payroll data through June 2026. Employment of workers aged 22 to 25 in the most AI-exposed occupations, such as software development, now stands 19% below where it would be had it kept pace with less-exposed jobs. Experienced workers show no comparable gap. The gap comes mostly from less hiring, not more firing, and it is concentrated in jobs where AI substitutes for tasks rather than assisting with them. It persists when tech companies and computer occupations are excluded.
The authors are also careful about what this does not prove. They call their findings “early, descriptive indicators” rather than causal estimates. The pattern weakens when they control for education, shows some divergence that predates generative AI, and is stronger in the ADP sample than in national surveys.
The strongest counterpoint comes from the New York Fed. In a June 2026 analysis, economists Natalia Emanuel, Emma Harrington and Amanda Pallais estimate that remote work explains 64% of the rise in unemployment among young college graduates between 2017-2019 and 2022-2024. Their logic is simple: it is harder to train a beginner over video, so firms with distributed teams hire more experienced people. Software engineering is their example of a “remotable” job. They also note that the rise in youth unemployment came before AI spread widely.
Both can be true. Remote work may explain the first years. AI may matter more from here on. The honest answer today: AI is a plausible and growing factor, not a proven main cause.
3. Visas and international students
For graduate programs, this is the clearest factor. According to the Taulbee Survey, nonresidents make up 53% of computer science master’s enrollment, and 63% of newly admitted master’s students come from outside North America. Total master’s enrollment in those departments fell about 26% in 2025. In the spring 2026 graduate survey of computing departments, 86% said visa delays and denials hurt their incoming class.
The wider numbers point the same way. The Institute of International Education’s Fall 2025 Snapshot found new international enrollment down 17% and international graduate students down 12%. The Clearinghouse counted 5.9% fewer international graduate students in fall 2025. International first-year applicants on the Common App fell 10% in the 2025-26 cycle, according to its end-of-season report. In September 2025, the President also ordered a $100,000 payment for many new H-1B work-visa petitions for people outside the U.S., and in September 2026 extended that restriction for another 12 months. For a foreign student weighing a U.S. degree, the job at the end looks harder to reach.
That fits the graduate curve: the steepest drop (18.6%) came in fall 2025, when visa problems hit, not this year. The Clearinghouse’s preliminary report does not break out international students. Its January report will.
4. Students moving to neighbouring programs
Not every student who leaves computer science leaves computing. In the fall 2025 survey, departments most often reported declines in computer science, software engineering and information systems, and growth in computer engineering, data science, cybersecurity and AI. Some leaders said students want a more “physical” major that feels less exposed to AI. At the graduate level, AI master’s programs were the only computing area with average growth (7%).
Some of that movement leaves the “computer science” family in federal statistics: computer engineering is classified under engineering, and data science has its own code. My reading: part of the 12.4% is students moving next door, not talent lost to the field. How large a part, the current data cannot say.
5. Cost and demographics
Holsapple points to affordability as one reason students are choosing community colleges. But tuition and demographics affect every major. They do not explain why computer science fell while engineering and health rose at the same schools. My reading: cost shapes where students enroll, not why they are leaving this particular field.
The verdict
The decline is real, large and broad. Its causes are layered: a hiring slump that started before generative AI, a sharp visa-driven fall in international graduate students, a shift to adjacent programs, and, on top, a growing fear that AI will do the work juniors used to do. That fear is reasonable. The evidence that AI already caused most of the damage is not there yet.
One more nuance. Bloomberg cites an August report from the Bureau of Labor Statistics as finding that computer and mathematics jobs would feel the most impact from AI. BLS did publish new AI exposure categories on August 27, 2026, but it says explicitly that “exposure does not imply job loss.” In its 2025-2035 projections, computer and mathematical occupations are the fifth fastest-growing major group (+7.3%). Software developers are projected to grow 10%. Computer programmers, the narrower role of writing code to specification, are projected to shrink 7%, and BLS names AI as a reason.
What this means for companies
- Plan for a smaller pipeline. This year’s record graduating classes come from the boom. New majors fell 13% in 2025, so by my estimate, smaller graduating classes arrive from about 2029. Companies that stop hiring juniors now will compete for a smaller pool of mid-level engineers later.
- Redesign the junior role, don’t delete it. IBM says it plans to triple U.S. entry-level hiring in 2026, and its HR chief told an audience that companies must “rewrite every job” (IBM Think, March 2026). That is IBM’s own plan, not an independent result. The idea behind it holds: if AI writes more of the first draft, juniors can spend more time on reviewing, testing, talking to users and learning the system.
- Train in person where it matters. If the New York Fed is right, distance is part of the problem. Pair juniors with seniors in the same room for at least part of the week, and treat code review as teaching.
- Measure outcomes, not output. If AI writes more code, the scarce skill is checking it. See the referee shortage and why Amazon told engineers to stop tokenmaxxing.
- Budget for international talent. Visa costs and delays are now a planning item, not an HR footnote.
I wrote earlier about the risk behind all this: the bottom rung. The danger is not the lost junior jobs. It is the missing layer that later becomes your seniors.
What this means for students, parents and career switchers
- Look at the full picture before switching. In the New York Fed data, recent computer science graduates had higher unemployment (7.0%) than the average graduate (4.2%), but a much higher early-career median wage ($87,000 against $58,000) and far less underemployment (19% against 39%). A computer science degree is riskier to start with than it used to be. It is not a bad degree.
- Build skills that AI makes more valuable. Reading and testing code you did not write. Understanding how a whole system fits together. Security basics. Explaining trade-offs to people who are not engineers. Practising with AI coding tools is part of this, but so is knowing when their output is wrong. As GitHub’s latest data show, the bottleneck has moved to review.
- Add a domain. Software plus health, energy, manufacturing, finance or public services is harder to replace than generic coding, because the hard part is knowing what to build.
- Ask departments for their own numbers. Parents and students can ask: what share of last year’s graduates had a job in the field six months after graduation? How many had internships? The answers vary widely between schools.
- For career switchers: the “learn to code in 12 weeks” route into a generic junior developer job is the hardest path right now. Combining your existing domain knowledge with software skills is a stronger one.
- Don’t decide on one headline. Students who start now graduate around 2030, into a market that will look different again. Fewer graduates may also mean less competition then. That is my reading, not a forecast.
Honest limits
- Preliminary data. The fall 2026 figures cover 46% of institutions, are unweighted and will be revised in January 2027. The prior-year comparison (−6.9%) comes from this year’s panel and differs from last year’s final figure (−8.1%).
- No international breakdown yet. The visa explanation for graduate programs rests on other sources (CRA, IIE, the Clearinghouse’s fall 2025 data), not on this report.
- Survey limits. The Taulbee Survey covers PhD-granting departments only. The pulse surveys reflect department leaders’ views and estimates.
- Postings are not hires. Indeed counts job ads, not jobs filled.
- Lagged outcomes. The New York Fed’s by-major data describe 2024.
- Competing studies. The Stanford and New York Fed analyses use different data and methods. Neither is a clean causal test.
- My own arithmetic (the roughly 530,000 estimate, “about three quarters”, “a fifth below” and “two-thirds below the peak”) is rounded and marked as such.
- Not covered: I could not read some campus-level reports in full, so I left them out.
Sources
- National Student Clearinghouse Research Center, Preliminary Fall Enrollment Trends, Fall 2026 (dashboard, Figure 3.5) and methodological notes, October 29, 2026
- National Student Clearinghouse, press release, October 29, 2026
- National Student Clearinghouse Research Center, Final Fall Enrollment Trends 2025 and major fields data appendix, January 15, 2026
- Bloomberg, via Business Standard, October 29, 2026 (secondary coverage, quote from Matthew Holsapple)
- Computing Research Association, 2025 Taulbee Survey results, June 2026
- CRA CERP, undergraduate pulse survey, October 2025, and graduate pulse survey, April 2026
- Indeed Hiring Lab, job postings tracker, Software Development (CC BY 4.0; also on FRED), retrieved October 29, 2026 (data through October 23, 2026)
- Indeed Hiring Lab, The Labor Market Is Tilting Toward Seniority, July 23, 2026
- Federal Reserve Bank of New York, The Labor Market for Recent College Graduates (by-major data released February 4, 2026; rates updated August 6, 2026)
- Brynjolfsson, Chandar, Chen, Canaries in the Coal Mine?, Stanford Digital Economy Lab, August 2026
- Emanuel, Harrington, Pallais, Remote Work Leaves Younger Workers Sidelined, Liberty Street Economics, June 1, 2026
- U.S. Bureau of Labor Statistics, Employment Projections 2025-2035, AI exposure categories, Software Developers and Computer Programmers, August 27, 2026
- Institute of International Education, Fall 2025 Snapshot on International Student Enrollment, November 2025
- Common App, End-of-season report 2025-2026, August 20, 2026
- Proclamation 11069, Restriction on Entry of Certain Nonimmigrant Workers, Federal Register, September 23, 2026
- IBM Think, The bottom rung returns as AI reshapes entry-level jobs, March 2, 2026
Researched and drafted with AI assistance, including Anthropic’s Claude.
