Computer science degrees more than doubled in a decade. Then recent-graduate unemployment jumped across three computing fields at once, including the largest increase of any field of study in computer engineering. Here is what the newest federal and survey data show, and what it should change about how families talk about the major.
In spring 2026 the National Student Clearinghouse Research Center reported that undergraduate enrollment in computer and information sciences fell 8.4 percent at four-year institutions and 11.2 percent at two-year colleges. Overall undergraduate enrollment grew 1.3 percent over the same period. College enrollment grew. Enrollment in this major fell.
That decline arrived about two years after the deterioration became visible in the recent-graduate labor data. The Clearinghouse does not report why students moved, and the timing is consistent with several explanations, but it is the first pullback large enough to see. By the time a high school junior hears that computer science hiring has cooled, the graduates who enrolled during the boom have already been looking for work for a year or two. The gap between what the enrollment numbers say and what the hiring numbers say is the whole story of this piece, and it is worth walking through in the order the data actually arrived.
A Major That Doubled
Figure 1. Bachelor’s degrees conferred in computer and information sciences, 2010-11 through 2021-22. Source: NCES Digest of Education Statistics, table 325.35, the last year that table covers. Provisional federal completions data tabulated by the National Science Board put the 2023-24 figure at 121,836 on a slightly different field definition. The boxed enrollment figures are a separate measure from a separate source, National Student Clearinghouse Research Center, spring 2026. Analysis by College Transitions.
Between 2010-11 and 2021-22, American colleges went from conferring 43,066 bachelor’s degrees in computer and information sciences to conferring 108,503. That is an increase of 151.9 percent in eleven years.
Nothing about the curve is gradual in the second half. Annual production rose by about 44,000 degrees between 2015-16 and 2021-22 alone, which is to say the field added more graduating capacity in six years than it had in total in 2010-11. Departments expanded, introductory sequences were capped, and a generation of families came to treat the major as the safest quantitative bet available to a student who was good at math and did not want medical school.
The series above runs through 2021-22, the last year the NCES Digest table covers. The expansion did not stop there. The National Science Board’s 2026 science and engineering report, drawing on a special tabulation of provisional federal completions data, puts computer sciences bachelor’s degrees at 121,836 in 2023-24. That figure uses a slightly different field grouping than the Digest table and is provisional, so it does not belong on the same plotted line, but the direction is not in doubt. Degree production was still climbing into 2024, which is the year the recent-graduate labor-market data in the next section describe.
Degree counts are a lagging measure of student demand, and that lag is the reason the spring 2026 enrollment decline should not be read as the end of the problem. Students graduating between now and the end of the decade generally picked their majors while computing enrollment was still expanding. The 2026 pullback will feed into bachelor’s degree production only gradually over the next several years. Most near-term graduating cohorts entered college before it.
The Hiring Side Went the Other Way
Figure 2. Unemployment rate for recent graduates aged 22 to 27, 2022 to 2024, showing the 13 largest increases, the all-field average and the 5 largest declines of 73 fields. Source: Opportunity Data analysis of Census Bureau American Community Survey 1-year microdata. Analysis by College Transitions.
The clearest read on what happened to those graduates comes from Census Bureau American Community Survey microdata, analyzed by field of study for graduates aged 22 to 27 who hold a bachelor’s degree. Between 2022 and 2024, unemployment among recent graduates rose 0.7 percentage points across all fields, from 3.6 percent to 4.3 percent. That is a modest, economy-wide drift.
Computing did not drift.
Computer engineering went from 2.3 percent unemployment to 7.9 percent, an increase of 5.6 percentage points and the largest of any of the 73 fields in the series. Information systems and management went from 2.7 percent to 7.0 percent, an increase of 4.3 points and the fourth largest. Computer science itself went from 4.4 percent to 6.7 percent, an increase of 2.3 points, which ranks 13th of 73.
The size of that swing is easier to feel in context. In 2022, computer engineering graduates were unemployed at 2.3 percent against an all-field rate of 3.6 percent. By 2024, at 7.9 percent, the field was tied with anthropology for the second-highest unemployment rate in the series, behind only fine arts. Liberal arts, meanwhile, went the other way, from 7.9 percent down to 4.0 percent.
The fields that improved over the same period cut against the intuition most families carry into these conversations. Aerospace engineering fell 6.0 points. Liberal arts fell 3.9. History fell 3.4. None of those movements should be over-read either, for reasons covered below, but the tidy picture of technical majors rising while humanities majors sink does not survive contact with this particular two-year window.
Three Fields Moving Together
Any single-year, single-field estimate from a household survey deserves suspicion. The organization publishing this panel says so directly, warning that the most eye-catching one-year jumps in an individual field usually sit inside the margin of sampling error.
So the question is not whether computer engineering’s 5.6-point jump could be noise. It could. The question is whether three related fields, estimated separately from the same survey, all moved the same direction at the same time by amounts that large.
Here is the full six-year panel for the three computing fields.
| Field and measure | 2018 | 2019 | 2021 | 2022 | 2023 | 2024 |
| Computer Science, unemployed | 6.3 | 5.2 | 5.2 | 4.4 | 6.2 | 6.7 |
| Computer Engineering, unemployed | 2.2 | 2.9 | 3.8 | 2.3 | 7.6 | 7.9 |
| Information Systems, unemployed | 3.5 | 5.8 | 4.7 | 2.7 | 5.0 | 7.0 |
| Computer Science, underemployed | 23.7 | 22.9 | 22.7 | 20.9 | 22.8 | 23.4 |
| Computer Engineering, underemployed | 23.1 | 17.6 | 20.3 | 17.6 | 20.0 | 18.8 |
| Information Systems, underemployed | 45.0 | 44.9 | 42.5 | 43.6 | 42.6 | 45.9 |
Table 1. Unemployment and underemployment rates, recent graduates aged 22 to 27 holding a bachelor’s degree, percent. Source: Opportunity Data analysis of Census Bureau American Community Survey 1-year microdata. Analysis by College Transitions.
There is no 2020 column because the Census Bureau never released standard one-year estimates for 2020, having collected about two thirds of its usual responses during the pandemic and found the respondents meaningfully different from the people it missed.
The panel stops at 2024 for a reason worth knowing, because it will hold for every article published on this subject for a while. As of August 2026 the Census Bureau had not set a release date for the 2025 one-year estimates, citing a new Commerce Department order on disclosure avoidance that it is still working to comply with. Until that file appears, 2024 is the most recent year anyone can report for unemployment by field of study.
The 2022 figures were unusually good, which inflates the size of the change. Reading the panel from 2018 instead of 2022 gives a more conservative picture: computer science unemployment ran between 5.2 and 6.3 percent in the pre-pandemic years and sits at 6.7 percent now, while computer engineering ran between 2.2 and 3.8 percent in 2018, 2019 and 2021 and sits at 7.9 percent. Computer engineering is the field that has changed most against its own history, not just against its 2022 low.
On sample size, these are not thin cells. The 2024 estimates rest on unweighted samples of roughly 2,300 computer science graduates, 1,100 in information systems, and 560 in computer engineering. Even so, the approximate standard errors available for a series like this one work as a screen rather than a formal variance estimate built from the survey’s replicate weights, and no single field’s two-year change should be treated as settled on its own.
What makes the pattern persuasive is coherence rather than statistical independence. The three fields share an economy and a labor market, so their moving together amounts to one signal observed three times rather than three separate confirmations of the same finding. Against a background in which the all-field rate moved 0.7 points, that is considerably harder to write off as sampling noise than any one of the three would be alone.
The underemployment column in the panel is the steadier of the two measures, and it barely moved. Computer science ran between 20.9 and 23.7 percent across all six years. Computer engineering ran between 17.6 and 23.1 percent. Whatever changed between 2022 and 2024 shows up in whether computing graduates were employed at all, not in the degree-level character of the jobs held by those who were.
There is corroboration outside the survey, and it runs closer to the present. Indeed’s Hiring Lab tracks job postings by occupational category against a February 2020 baseline. As of January 23, 2026, software development postings stood 31.8 percent below that baseline, data and analytics 40.1 percent below, IT infrastructure and support 35.2 percent below, and IT systems and solutions 32.2 percent below. Every tech category the lab tracks was at least 30 percent under its pre-pandemic level while total US postings sat 4.4 percent above it.
One piece of that is better news than the rest. Software development postings were flat over the year ending January 23, 2026, after several years of decline. Postings are not hires, and a proprietary index is not an official statistic, but a category that has stopped falling is a different situation from one still in descent.
This Isn’t Mainly a Barista Problem
Figure 3. Unemployment against underemployment by field of study, recent graduates aged 22 to 27, 2024. The chart separates difficulty getting hired from the degree level of the job once hired, two measures that are routinely conflated. Underemployment is the share working in jobs that typically do not require a bachelor’s degree. Source: Opportunity Data analysis of Census Bureau American Community Survey 1-year microdata. Analysis by College Transitions.
This is the finding that should change how the conversation goes at the kitchen table, and it is the one almost nobody reports.
Underemployment, defined here as the share of employed graduates working in jobs that typically do not require a bachelor’s degree, runs at 41.9 percent across all fields for recent graduates. Computer science sits at 23.4 percent. Computer engineering sits at 18.8 percent. Both are among the lowest figures in the entire series.
So computing graduates in 2024 were harder to hire than average and far more likely than average to be doing degree-level work once hired. Those two facts sit in tension with the story that gets told about the major, which usually treats a bad job market as evidence that a degree has stopped being worth anything.
The scatter above separates those two ideas. The horizontal axis is the difficulty of getting hired at all. The vertical axis is the quality of the job on the other side. Fine arts sits in the upper right, with 9.2 percent unemployment and 64.0 percent underemployment, which is the genuinely difficult quadrant. Nursing sits in the lower left, at 2.2 percent unemployment and 13.2 percent underemployment, which is what a tightly licensed pipeline into a specific occupation looks like. Computer science and computer engineering sit in the lower right, which is the corner almost nobody occupies. High unemployment usually travels with high underemployment, because fields that struggle to place graduates tend to place them badly. Computing broke that pattern in 2024. Fewer recent computing graduates were employed, and the ones who were employed remained unusually likely to be doing work that typically requires a bachelor’s degree.
Information systems and management is the exception within computing, and families should know the difference. It carries computing-level unemployment at 7.0 percent and near-average underemployment at 45.9 percent. Students sometimes treat the two as interchangeable options. The early-career profiles here are very different.
What These Numbers Do Not Establish
A reader could take the three figures above and assemble a confident theory about artificial intelligence eliminating junior engineering roles. The data here do not support that, and anyone who tells you they do is filling in a gap.
What the data show is a rise in unemployment among recent computing graduates concentrated in a two-year window, alongside a collapse in technology job postings. It does not identify a cause. Interest rates, the end of pandemic-era hiring, layoffs concentrated at large technology employers, a surge in the number of candidates produced by the enrollment boom itself, and changes in how entry-level work is assigned are all live explanations, and several of them can be true at once. Sorting them would require data that does not exist yet.
The people who publish the postings index say something similar about their own numbers. Laura Ullrich, director of economic research at Indeed’s Hiring Lab, put it this way: “AI didn’t cause the bust in hiring in the tech sector, but it might be preventing it from recovering at the same rate it would have.” The available evidence does not let anyone decompose the computing slowdown into artificial intelligence, post-pandemic normalization, interest rates, layoffs and candidate supply, and a piece that claims to have done so is guessing.
The degree counts are administrative records and can be taken at face value. The Clearinghouse reports that institutions submitting enrollment data to it account for 97 percent of all enrollments at Title IV degree-granting institutions, which makes those figures close to a census. The unemployment and underemployment figures come from a household survey and carry sampling error, which is why this piece leans on three fields moving together rather than on any single number. The postings index is a private measure of one company’s marketplace.
Families who go looking will find slightly different numbers elsewhere, and should not be alarmed by that. The Federal Reserve Bank of New York publishes its own analysis of the same 2024 survey year and reports computer science at 7.0 percent unemployment and computer engineering at 7.8 percent, against the 6.7 and 7.9 used here. Its overall underemployment figure, 42 percent, matches this panel’s 41.9 almost exactly. Different teams make different sample and classification choices. This piece uses one source throughout rather than mixing them, which is why its all-field comparison figure is 4.3 percent rather than the higher headline rate the New York Fed reports from a different survey.
One more limit is worth stating plainly. Every figure here describes recent graduates, aged 22 to 27. None of it describes what happens to a computer science graduate at 32, and the mid-career earnings advantage in computing has historically been large. A rough entry into the labor market and a bad forty-year outcome are different claims.
What a Family Should Actually Do With This
The recommendation is not to avoid the major. It is to stop treating it as insurance.
For most of the past fifteen years, computer science functioned in family conversations as the answer to the question of what a smart, undecided student should study in order to be safe. The recent data make that safety premium much harder to defend. A student who chooses it now should choose it because they want to build software, not because it is the thing that guarantees a job, and counselors should be comfortable saying that out loud.
Three things follow from the numbers.
Students who are already in the major should not switch out on the strength of a two-year window. The underemployment figures say that graduates who do land jobs are landing real ones, and the student pipeline has begun shrinking, which could eventually ease some supply pressure if that decline persists and employer demand stabilizes. Degree production was still rising as recently as 2023-24, so relief on the supply side is not imminent.
Students choosing between computer science and information systems should understand that these are not the same bet. The 2024 gap in underemployment between the two, 23.4 percent against 45.9 percent, is 22.5 percentage points. On that measure, information systems sits closer to philosophy at 45.8 percent and psychology at 48.5 percent than it does to computer science.
Students whose main goal is occupational security should notice what the fields on the left side of Figure 3 have in common. Most of them run a licensed or credentialed pipeline into a defined occupation. Nursing is the clearest case, at 2.2 percent unemployment and 13.2 percent underemployment. That structure carries real constraints, including licensure, clinical placement and a narrower set of exits, and it is not a good fit for every student who is good at math. What it produces is a tighter connection between the degree and the job than a general technical major supplies. A family that wants that connection should go looking for it directly rather than assuming computing comes with it.
For counselors, the practical version of all this fits on an index card. The major is still a strong one. It is no longer the risk-free one. On recent-graduate unemployment, 2024 is the weakest point in this six-year computing panel, and the students most exposed to it are the ones who picked computing because someone told them it was safe rather than because they liked the work.
The spring 2026 enrollment decline is consistent with students beginning to make that adjustment on their own, roughly two years after the labor-market deterioration became visible. The families who read the data earlier get to make the choice deliberately instead of following the crowd out the door.
Data vintages. Degrees conferred through 2021-22 (NCES Digest of Education Statistics, table 325.35, the last year that table covers), with a provisional 2023-24 figure from the National Science Board, The State of U.S. Science and Engineering 2026, special tabulation of the IPEDS Completions Survey. Enrollment through spring 2026 (National Student Clearinghouse Research Center). Unemployment and underemployment by field of study through 2024, the most recent year available and likely to remain so while the 2025 American Community Survey one-year release is pending (Opportunity Data analysis of Census Bureau American Community Survey 1-year microdata, graduates aged 22 to 27 holding a bachelor’s degree). Job postings through January 23, 2026 (Indeed Hiring Lab, US Labor Market Quarterly Verticals for Q4 2025, published February 20, 2026). Cross-reference figures from the Federal Reserve Bank of New York, The Labor Market for Recent College Graduates, February 2026 update.
