From Sanskriti Khandelwal | Product & Market Analysis
Entry-Level Jobs in 2026: Why Hiring Is Up 5.6% and Graduates Still Struggle
On this page
Two numbers, both 5.6%, describe the class of 2026. NACE says the median employer planned to hire 5.6% more new graduates than last year. The New York Fed says unemployment among recent graduates sat at about 5.6% through mid 2026. Entry-level jobs are not vanishing, and they are not easy to land. Both claims hold, because each source counts something different.
Key takeaways
- NACE's 5.6% is a median of hiring plans, not a count of jobs. It comes from 185 employers surveyed in February and March 2026, and the comparable fall 2025 median was minus 2.4%.
- Roughly a quarter to a third of employers say AI has already cut entry-level need. NACE's own respondents and a separate poll of 1,000 hiring managers land in that same band, which is the strongest agreement in this whole debate.
- Entry-level postings on Indeed fell 7.5% in the year to May 2026 while senior postings rose 14.7%. The squeeze shows up in who gets advertised for, not in mass layoffs.
- Graduate unemployment is high and flat, not spiking. An NBER study of summer 2026 found no significant rise against older graduates or young workers without degrees, which argues against an AI shock so far.
The short answer
Entry-level hiring for graduates is not collapsing, and it is not healthy. Large employers planned modest increases for the class of 2026, while overall entry-level postings shrank and senior postings grew. Graduate unemployment is elevated but stable. AI is cutting some entry-level tasks, yet the measured job losses are still small and concentrated.
Entry-level jobs data that looks like a contradiction
Read the headlines from spring 2026 in sequence and you get whiplash. In April, the National Association of Colleges and Employers reported that hiring for the class of 2026 was up. In July, a survey of 1,000 hiring managers reported that nearly half would rather buy AI tools than train a graduate.
Both reports are real. Both are correctly quoted in most coverage. The confusion comes from treating them as measurements of the same quantity, when they are not.
This post lines up five sources that matter for entry-level jobs in 2026. For each one, it states what was counted, who was asked, and when. Then it shows how the numbers fit together, and which entry-level roles are actually growing or shrinking.
| Source | What it counts | Sample and window | Headline |
|---|---|---|---|
| NACE Job Outlook Spring Update | Employer hiring plans, median change | 185 employers, 12 Feb to 17 Mar 2026 | +5.6% planned graduate hires |
| ResumeTemplates.com survey | Hiring manager intentions and attitudes | 1,000 US managers, firms with 101+ staff, May 2026 | 48% prefer AI tools over training a graduate |
| New York Fed | Unemployment and underemployment, ages 22 to 27 | National survey data, through Q2 2026 | About 5.6% unemployed, 42% underemployed |
| Indeed Hiring Lab | Job postings by seniority level | All US postings on Indeed, to May 2026 | Entry-level postings down 7.5% year on year |
| Stanford Digital Economy Lab | Payroll employment by age and AI exposure | ADP payroll records, through June 2026 | 19% gap for ages 22 to 25 in exposed jobs |
Plans, attitudes, unemployment, postings and payroll are five different objects. A rising plan and a falling posting count can both be accurate in the same quarter.
What the NACE job outlook actually measured
NACE has surveyed employers about graduate hiring for decades, and its job outlook is the most quoted number on the class of 2026. It deserves a careful read rather than a headline read.
The Spring Update reports that employers planned to hire 5.6% more class of 2026 graduates than class of 2025. That figure is a median. NACE switched to the median in spring to reduce distortion from outliers.
The fall survey had been reported as a mean of 1.6%. NACE's own footnote says the same fall data expressed as a median was minus 2.4%. So the like-for-like swing is from minus 2.4% to plus 5.6%, which is larger than most coverage implied.
I would not quote the 5.6% without the word "median" attached. It describes the plan of the middle employer in a modest sample. It does not tell you how many graduate jobs exist.
The spring sample was 185 respondents: 142 NACE employer members and 43 nonmembers. The member response rate was 19.9%. NACE members skew toward large employers that run structured campus recruiting programmes.
That skew shows in the detail. Employers with more than 5,000 staff planned to increase hiring by 8.7%. More than one-third of all respondents planned more hires, against one-quarter in the fall. Company growth and succession planning were the main reasons given.
What NACE found on AI
NACE also asked about AI, and its April 2026 analysis is more nuanced than the hiring headline. Employers said more than one-third of their entry-level jobs now require AI skills, nearly triple the share they reported in fall 2025.
On replacement, the answers were split. More than half said AI is not reducing entry-level tasks. Just over one-quarter said AI has reduced the need for those tasks. Only 11% said they were discussing how AI might replace some positions.
What hiring managers say about AI and entry-level jobs
The survey that produced the "AI over graduates" headlines was commissioned by ResumeTemplates.com and run on the Pollfish panel in May 2026. It polled 1,000 US managers responsible for entry-level hiring at companies with at least 101 employees. The stated margin of error is plus or minus 3.1 points.
It is a commercial, opt-in survey, so it sits below NACE and the Fed in any sensible source ranking. It is still worth reading in full, because its full answer set says something different from its headline.
According to the published report, 48% of managers would rather invest in AI tools than hire and train a recent graduate. 55% said their company had shifted at least part of its entry-level budget to AI. 45% said they had restructured so one senior worker plus AI does the work of several entry-level hires.
Those are attitudes and reported reorganisations. They are not payroll records.
The numbers that did not make the headlines
The same survey asked how many class of 2026 graduates each company would hire. 48% said about the same number as last year. 17% said more, 18% said fewer, 5% said none and 12% were undecided. 74% planned to keep starting pay flat.
So the modal hiring manager in this poll was holding graduate headcount steady. The report itself says the market is not closing to new graduates. What it shows is a tilt, not a cliff.
The most telling figure is 30%, the share saying AI has already reduced their need for entry-level graduate hires. NACE's respondents put the same figure at just over one-quarter. Two surveys with different samples, sponsors and questions land within a few points of each other.
Recent graduate unemployment is high, but not spiking
The New York Fed's tracker covers graduates aged 22 to 27 with a bachelor's degree or higher. Its 2026 Q2 update says conditions remained challenging. Unemployment stayed elevated at about 5.6%, and underemployment edged up to 42%.
Underemployment means working in a job that does not usually require a degree. At 42%, that is more than 4 in 10 recent graduates. For many readers, that number describes the class of 2026 more accurately than the unemployment rate does.
An unemployment rate is a stock. It counts people who are looking and not working at a point in time. It barely moves if hiring slows but layoffs stay low, because fewer people enter unemployment and fewer leave it. Economists call this a low-hire, low-fire market.
In that kind of market, the pain shows up as longer searches and more graduates taking non-graduate jobs. Both are consistent with a stable 5.6% and a rising 42%.
The best test of an AI shock so far
Robert Fairlie and Jane Wu tested the AI explanation directly in an NBER working paper issued in September 2026. Using Current Population Survey microdata for June to August 2026, they compared recent graduates with older graduates and with young workers without degrees.
They found that unemployment "did not spike in summer 2026 relative to summer months in previous years." It also did not rise significantly against either comparison group. Counting people who want a job but are not searching adds nearly 2 points to recent graduate unemployment, and still produced no significant 2026 increase.
That is a strong result against the idea that AI has already displaced graduates at scale. It does not rule out slower hiring into specific roles, which is a different question.
Job postings show entry-level jobs thinning, not collapsing
Postings measure what employers are advertising for right now. They lead unemployment data and they lag nothing, which makes them the most current signal in this set.
Indeed Hiring Lab's July 2026 analysis classified every US posting as entry, mid or senior level, based on the tasks described rather than the job title. As of May 2026, senior postings were up 14.7% year on year. Entry-level postings were down 7.5% year on year and had been trending down since a 2022 peak.
Indexed to January 2025, senior postings were up 13.5%, mid-level down 6.7% and entry level down 6.3%. Entry-level roles still made up about 46% of all postings, against 14% senior.
The payroll evidence
Payroll data adds the age dimension that postings lack. The Stanford Digital Economy Lab's Canaries in the Coal Mine paper, revised in August 2026, uses ADP payroll records through June 2026. It finds employment of 22 to 25 year olds in the most AI-exposed occupations is 19% below where it would be had it kept pace with less-exposed peers.
The authors say the gap comes mainly from less hiring rather than more separations. They also say it is concentrated where AI substitutes for tasks, while employment is flat or rising where AI complements workers. The software slice of this is covered in depth in the analysis of the junior developer pipeline, so it is not repeated here.
How all four entry-level hiring signals hold at once
Put the sources together and the contradiction disappears. Each measures a different slice of the same slow-moving shift.
NACE speaks for large firms with formal graduate programmes. Those firms hire for succession, and they planned increases. The manager poll covers a much wider range of firm sizes, and it shows fewer expanders. Indeed covers every advertised role, most of which are not graduate jobs at all.
Plans come first, postings second, hires third and unemployment last. A firm can plan to hire 10 graduates, advertise fewer general entry roles, and fill the graduate slots from interns it already knows. NACE's spring data showed intern hiring plans up nearly 4%, which fits that pattern.
The AI effect is real and still narrow
This is the part I am most confident about. Two independent surveys say roughly 25% to 30% of employers have already cut entry-level need because of AI. Payroll data finds a gap in exposed occupations. Yet economy-wide unemployment for graduates has not moved. Those facts are compatible if AI is reshaping a minority of employers and roles, not the whole graduate market.
My read is that the hiring-manager headlines overstate the speed and NACE's headline understates the strain. The honest summary is a market that is open, smaller at the bottom of white-collar ladders, and tilted toward experience. The broader corporate pattern behind that tilt is examined in the piece on AI layoffs and the overhiring correction.
Which entry-level roles are growing and which are shrinking
Averages hide most of what a graduate or a hiring manager needs. The sources do agree on some direction by sector, even where they measure differently.
| Signal | Growing or holding | Shrinking or weak |
|---|---|---|
| NACE employer hiring plans, spring 2026 | Information, engineering services, wholesale trade, construction, professional services | Utilities, computer and electronics manufacturing, food and beverage manufacturing, pharmaceutical manufacturing |
| Indeed entry-level share of postings, 2019 to 2025 | Retail, up 4.4 points | Healthcare, down 5.2 points |
| Indeed entry-level share, Q1 2026 level | Personal care and home health, 91.3% entry level | Software development, 4.5% entry level |
| Stanford payroll data, 2026 | Roles where AI complements the worker | Roles where AI substitutes for core tasks |
NACE's industry lists come from its Job Outlook summary page and carry no per-industry percentages. Indeed's sector shares include non-graduate roles.
Engineering services, construction and professional services appear on NACE's growth list, and they share a trait. Much of the entry-level work involves sites, clients or regulated sign-off, which AI does not do on its own. Information appears on the growth list too, which sits awkwardly with Indeed's tiny entry share in software. My guess is that large information firms are hiring graduates into programmes, while open entry postings in software stay scarce.
Where the ladder is thinner
Software development is the clearest weak spot, with just 4.5% of postings at entry level in Q1 2026. In tech, Indeed found most of the shift came from mid-level postings falling, with senior roles growing by 9 points. Manufacturing segments on NACE's decline list look more like a cyclical story than an AI one.
The skill bar is moving in every sector. NACE found 28% of employers seeking early-career talent who can use AI at work. How those signals get tested in interviews is covered in the guide to AI fluency interview questions.
Where this argument is weakest
A reconciliation is only as good as its weakest input, and several inputs here are soft.
NACE's spring update rests on 185 employers with a member response rate under 20%. The manager poll is an opt-in commercial panel run by a company that sells resume templates. Both measure stated intentions, which often differ from what firms actually do by autumn.
Definitions do not line up
Indeed's "entry level" includes home health aides and food service roles that never require a degree. The New York Fed counts graduates up to age 27 and includes advanced degrees. Stanford's group is 22 to 25 regardless of degree. The authors note their pattern weakens when controlling for education, which matters a lot for a post about graduates.
The case that AI is bigger than this
The strongest counterargument is timing. Unemployment is a lagging indicator, and the Fairlie and Wu window is only three summer months. If firms are quietly restructuring as the manager poll describes, the effect could appear in 2027 data rather than 2026. Stanford reports its gap has widened steadily since August 2025. I would weight that trend more heavily than any single survey, and I would revisit this post when the next two Fed updates land.
Frequently asked questions
Are entry-level jobs disappearing because of AI?
Not across the economy, based on 2026 data. Recent graduate unemployment held near 5.6% and an NBER study found no significant summer 2026 spike. AI is cutting entry-level need at some firms, with roughly a quarter to a third of employers saying so in two separate surveys. Payroll data shows a gap for young workers in AI-exposed occupations, driven mainly by slower hiring rather than layoffs.
What is the NACE job outlook for the class of 2026?
NACE's Job Outlook 2026 Spring Update, published in April 2026, found employers planned to hire 5.6% more class of 2026 graduates than class of 2025. That is a median from 185 respondents surveyed in February and March 2026. The comparable fall 2025 median was minus 2.4%. Employers with more than 5,000 staff planned an 8.7% increase.
What is the unemployment rate for recent college graduates in 2026?
The New York Fed puts unemployment for recent graduates aged 22 to 27 at about 5.6% through the second quarter of 2026. Underemployment, meaning graduates working in jobs that do not usually require a degree, edged up to 42%. The rate is elevated compared with the wider workforce but has stayed broadly flat during 2026.
Which entry-level jobs are growing in 2026?
NACE's spring survey found the most hiring increases in information, engineering services, wholesale trade, construction and miscellaneous professional services. Indeed data shows retail's entry-level share of postings rose between 2019 and 2025. Software development is the weakest area, with only 4.5% of postings at entry level in early 2026. Roles where AI complements workers are faring better than roles where it substitutes.
Why do hiring surveys and job posting data disagree?
They measure different things. Hiring surveys record what a sample of employers plan to do, often weighted toward large firms with graduate programmes. Posting data counts every advertised role, including many that need no degree. Unemployment data counts people still searching. Plans can rise while postings fall, especially when firms fill graduate places from intern pipelines instead of open adverts.
Do employers prefer AI over hiring new graduates?
Some do. A May 2026 poll of 1,000 US hiring managers found 48% would rather invest in AI tools than train a graduate, and 30% said AI had already cut their entry-level need. The same poll found 48% planned to hire the same number of graduates as last year and 17% planned more. Preference has not yet turned into widespread hiring cuts.
How to use this data this week
If you hire graduates, pull your own numbers before you read another survey. Count entry-level requisitions opened in the last 12 months against the prior 12, and note which ones were filled by former interns. If the count fell, write down whether AI or budget drove it. That single line will settle more internal arguments than any national headline.
If you are a graduate, aim at the sectors where both NACE and Indeed point the same way, and treat an internship as close to mandatory. 61% of managers in the May 2026 poll said they would consider a graduate without one only for certain roles or not at all. The skills that survive automation are examined in the piece on the tacit knowledge premium.
Related analysis
For the engineering version of this story, read why the junior developer pipeline is breaking. For how hiring screens are changing, see what technical interviews still signal.
References
- NACE, Employers expect to hire 5.6% more new college graduates this year, 15 April 2026. Used for the spring median, sample and large-employer figures.
- NACE, Hiring is flat for the college class of 2026, 12 November 2025. Used for the fall mean and the 25/60/15 split.
- NACE, Demand for AI skills in entry-level jobs nearly triples since fall 2025, 20 April 2026. Used for AI skills and AI replacement responses.
- ResumeTemplates.com, The class of 2026 hiring report, 2 July 2026. Commercial Pollfish survey, n=1,000. Used for hiring manager attitudes and plans.
- Federal Reserve Bank of New York, The Labor Market for Recent College Graduates, 2026 Q2 update. Used for unemployment and underemployment.
- Fairlie and Wu, The Early Impacts of AI on Employment among Recent College Graduates, NBER Working Paper 35796, September 2026.
- Indeed Hiring Lab, The labor market is tilting toward seniority, 23 July 2026. Used for postings by seniority and sector.
- Brynjolfsson, Chandar and Chen, Canaries in the Coal Mine?, Stanford Digital Economy Lab, revised 12 August 2026.
The weakest link in this source base is that the two employer surveys are small or self-selected and measure intentions, not hires. Figures are current as of 8 October 2026; the New York Fed updates quarterly.
Related reading