Big Tech quietly cut entry-level coding jobs by 65 percent since 2019. The vacancies never came back.
The single most important number in the AI labor debate this year is not a layoff statistic. It is a vacancy that was never posted. According to SignalFire's 2026 State of Tech Talent report, new-grad and entry-level hiring at the 12 Tech Majors has collapsed roughly 65 percent compared to 2019. At early-stage startups, the drop is 76 percent. The jobs did not move somewhere else. They stopped existing.
The consensus blames AI for killing junior jobs, but the real driver is a strategic reallocation of hiring budgets. Big Tech is hoarding senior talent to build AI systems that automate junior work. They are not replacing humans with bots. They are reshaping the workforce pyramid from the bottom, and the data proves the bottom is gone.
Three Datasets, One Conclusion
Total hiring at large tech companies sits 25 percent below the 2019 baseline. Engineering hiring, however, is down only 11 percent. Inside that shrinking pie, software engineers now account for 55 percent of all hiring, up from 46 percent in 2019. Big Tech did not stop hiring engineers. It stopped hiring inexperienced ones.
Stanford researchers revised their findings in August 2026 and found employment for 22-to-25-year-olds in highly AI-exposed occupations is about 19 percent below where it would be had it tracked less-exposed peers. That is up from a 15 percent shortfall in the July 2025 data vintage. The gap is widening.
An IZA discussion paper examining millions of job postings found a 14 to 15 percent relative decline in junior versus senior software developer vacancies after ChatGPT's release. The effect was isolated to software roles. It was absent in mechanical engineering. The researchers concluded that generative AI redefines entry-level work by raising the bar for what counts as a qualified junior hire.
The adjustment runs through hiring, not firing. Separations look normal. Nobody needed to fire anyone. The req just never got posted.
The Doors Stayed Shut
The tech industry has frozen junior hiring before. 2001 and 2008 both saw entry-level doors slam shut. Both times, the doors reopened when demand returned.
This time, the work itself changed.
Revelio Labs examined who is missing from payroll data and found the answer is almost entirely one cohort: workers between 22 and 25. Evan Sohn of Revelio Labs described the moment as the lowest hiring since 2021 and the lowest firing in a long time. Employers posted 7.27 million open jobs in July 2026, the highest reading since May. In that same month, total nonfarm payrolls declined by roughly 23,000 jobs.
The economy is not frozen. It is sorting. Companies are hiring. They are simply refusing to hire anyone under 26 for roles that AI now handles.
Stanford's earlier paper found a 13 percent relative decline in employment for early-career workers in AI-exposed occupations after controlling for industry-wide shocks like interest rate changes. The effect persisted through multiple data vintages. The Stanford and AI2Work researchers were explicit: the adjustment runs through hiring, not firing. This is not a layoff story. It is a non-hire story.
The junior coder hiring decline is not a cyclical pause. It is a structural redefinition of what entry-level work requires.
How the Bar Moved Up
The IZA paper explains the precise mechanism, and it is more insidious than a simple headcount reduction.
Rising experience requirements were driven primarily by employers asking for more experience within the same job titles. They did not rename the role. They did not shift hiring toward senior titles. They took the same junior software developer job and demanded more years of experience to qualify. The bar moved up while the title stayed the same.
The remaining junior vacancies shifted toward problem solving, communication, and attention to detail. Not AI-specific skills. Not prompt engineering. The market is not asking juniors to learn AI. It is asking them to be more expensive hires who can do what AI cannot.
Meanwhile, SignalFire's data shows each engineering manager at a Tech Major now manages roughly 12 engineers, up from 10. At startups, the ratio is about 15 engineers per manager. The system is optimizing for density of experienced output. Companies are paying senior engineers to build AI systems that handle the grunt work juniors used to do, then using the savings to hire more senior engineers. The flywheel feeds itself.
Top computer science graduates in 2025 are twice as likely to call themselves a founder compared to the class of 2022. They are 45 percent less likely to land a job at a Tech Major. Early-stage startups hired 7 percent more engineers in 2025 than in 2019, but those hires were not entry-level. The 76 percent collapse in startup junior hiring happened alongside a 7 percent increase in total engineering headcount. Startups are following the same playbook: hire experienced engineers, skip the training pipeline.
The Founder Factory
The structural outcome is already visible.
Step one: Big Tech used AI to absorb entry-level grunt work. Debugging, boilerplate generation, documentation, test writing. The tasks that once constituted a junior engineer's first two years now execute in seconds. The training pipeline that produced mid-level engineers was severed at the root.
Step two: The cohort that would normally be mid-level in two to three years is missing. They were never hired. They never accumulated the debugging scars, the code review reps, the production incident experience that turns a CS graduate into a competent engineer. The 22-to-25-year-olds who should be entering the pipeline now are instead calling themselves founders or working outside the traditional tech employment structure entirely.
Step three: Within 12 to 24 months, companies will face an acute shortage of engineers with three to five years of experience. The missing junior cohort cannot advance because it does not exist. The mid-level supply chain is broken, and the break happened three years ago. The lag effect is about to hit.
Step four: The market response will not be a sudden re-hiring of juniors. That door is structurally closed. The response will be a 20 to 30 percent premium on experienced developers as companies bid against each other for the shrinking pool of engineers who entered the workforce before 2022. Tech hiring overall sits at 75 percent of its pre-pandemic baseline. When demand returns, the supply of mid-level talent will not be there to meet it.
Step five: The adjustment mechanism will be a surge in AI-augmented bootcamps and alternative training pathways that bypass the traditional resume filter entirely. If Big Tech will not train juniors, the market will route around Big Tech. The fact that top CS grads are already twice as likely to identify as founders is the leading indicator. The salaried entry-level path is dead.
This is not a story about ambition spiking. It is a story about the alternative disappearing. The founder path is the new default, and that changes the nature of the companies being built. Founders born from desperation make different bets than founders born from opportunity. They take less venture capital because they cannot raise it. They build revenue-first because they have no safety net. They solve unglamorous problems because those are the only ones that pay immediately. The 2025 and 2026 CS graduating classes will produce a generation of companies optimized for survival, not scale, and that is a structural shift in the startup ecosystem that no accelerator or fund has priced in.
What would falsify this prediction? If Big Tech reopens entry-level hiring within 18 months, the founder factory dynamic dissipates. If the IZA and Stanford findings reverse in the next data vintage, the mechanism weakens. But the data is not moving in that direction. The gap is widening. The vacancies are not returning. The founder factory is not a theory. It is the only remaining door.
You Are Already Late
If you are an operator who can hire experienced engineers today, pay the premium now. In 18 months, the premium will be higher and the pool will be smaller. The market is mispricing the risk of a mid-level talent cliff because the data is only now becoming legible.
If you are a junior engineer waiting for Big Tech to reopen entry-level roles, the data suggests you will wait forever. The IZA finding that remaining junior vacancies demand problem solving and communication, not AI skills, points toward a single viable path: build something that proves you can operate at the level the market now demands. The traditional credential and internship pipeline no longer terminates in a job offer.
Companies must start building internal training programs that assume no one under 26 has been trained by a major employer in the last three years. The lost cohort is real. They are smart and they are unemployed through no fault of their own. The organizations that figure out how to convert them into productive engineers without the traditional apprenticeship structure will capture the talent that Big Tech abandoned.
For investors, the signal is clear. The founder factory dynamic means a surge of capable, desperate, technically literate founders entering the market with nothing to lose. Some of those startups will be built by people who would have been staff engineers at Google in a different world. The question is whether your fund is structured to find them before they stop looking for checks.
The Vacancies That Never Returned
Three years of missing entry-level hires means three years of missing mid-level engineers starting now. The market has maybe 18 months before the shortage becomes acute and the wage premium spikes. The adjustment will not be a reopening of junior roles. It will be a frantic scramble for anyone with three years of production experience and a pulse.
The vacancies never came back. The lost cohort is not coming back through traditional channels. The question is whether the market adapts fast enough to avoid a talent cliff that the data already predicts.
