The Software Engineer Shortage in 2026: What AI Changed

The short version

• New-grad and entry-level hiring is down roughly 65% at the largest tech companies and roughly 76% at early-stage startups compared with 2019, according to SignalFire's 2026 State of Tech Talent report (June 2026).
• Total engineering hiring held up far better than the headlines suggest: engineering hiring at the largest tech companies is down only 11% versus 2019, against a 25% decline in their overall hiring, and engineering hiring at early-stage startups is actually up 7%, according to SignalFire's 2026 State of Tech Talent report.
• Engineers now make up 55% of all hiring at the tech majors, up from 46% in 2019, according to SignalFire's 2026 State of Tech Talent report — engineering hiring shrank less than everything around it.
• Compensation moved up the seniority ladder, not down it: Levels.fyi's 2025 pay report puts median staff engineer total compensation at $457K, up 7.52% year over year, against $155K for entry level, up just 1.64% — though principal-level pay fell 6.58%, so the gain concentrates at staff level rather than at the very top.
• AI/ML engineering roles grew 39% as a share of hiring and forward deployed engineer roles grew 30%, both measured since ChatGPT's launch in 2022, according to SignalFire's 2026 State of Tech Talent report — the clearest signal of where scarcity relocated.
• The causal claim is contested. Stanford's SIEPR found unemployment rose almost identically for the most and least AI-exposed workers since 2022 (0.77 versus 0.85 percentage points), and Yale's Budget Lab finds no relationship between AI exposure and employment change at all.

Demand did not disappear, it changed shape

Start with the single cleanest measure. Indeed's software development job postings index, tracked by the St. Louis Fed, sat at 75.46 on July 31, 2026, against a February 2020 baseline of 100. Software development postings are roughly a quarter below where they were before the pandemic, and they have been flat-to-drifting for months rather than recovering.

That number is easy to misread as "fewer engineers needed." It is not what the employment data says.

The US Bureau of Labor Statistics still projects 15% growth for software developers, QA analysts, and testers from 2024 to 2034, from a 2024 base of 1,895,500 workers, with about 129,200 openings a year over the decade. That projection was last updated in August 2025, so treat it as the official baseline rather than a read on 2026 conditions — but it is not a forecast of contraction.

Two things are true at once. Openings are scarcer than they were, and the work has not gone away. What changed is who gets hired to do it.

The entry-level decline in numbers

The clearest data comes from SignalFire, which tracks career movement across hundreds of millions of professional profiles. Its 2026 report, published June 22, 2026, found:

• New-grad and entry-level hiring down roughly 65% at the tech majors versus 2019.
• New-grad hiring down roughly 76% at early-stage startups versus 2019.
• Graduates of the top 20 CS programs 45% less likely to take a role at a tech major than in prior years.
• Those graduates twice as likely to be a founder in 2025 as the 2022 class.

The academic work points the same direction. In "Canaries in the Coal Mine?" (November 2025), Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen used payroll records from the largest US payroll provider and found a 16% relative employment decline for workers aged 22 to 25 in AI-exposed occupations, even after controlling for firm-level shocks. Employment for experienced workers in those same occupations stayed stable or grew.

Their finding that declines concentrate in occupations where AI automates rather than augments is the mechanism that matters. Writing a well-specified CRUD endpoint is closer to automation. Deciding whether the endpoint should exist is closer to augmentation.

The bar did not rise because juniors got worse. It rose because the tasks that used to justify hiring a junior are now the tasks a model does first.

Where the scarcity moved

If the shortage inverted, the inverse should be visible in what is hard to hire and what it costs. It is.

CIO's June 2026 survey of the hardest IT roles to fill put AI/machine learning and cybersecurity tied at number one. Software engineering ranked sixth. Application development fell off the list entirely. The same CIO piece draws on the 2026 SANS/GIAC Cybersecurity Workforce Research Report — a cybersecurity-specific study, not a whole-workforce one — in which 60% of organizations named skills gaps rather than headcount shortages as their top workforce challenge, a 20-point lead over headcount that has widened from four points a year earlier.

Neal Sample, chief digital and technology officer at Best Buy, framed the problem in that report as "three skills, one person, small pool," and noted that "the center of gravity moved from people who build models to people who wield them." His most telling line: "Our most productive AI engineers in 2025 were not hired as AI engineers."

Compensation confirms it. Levels.fyi's 2025 end-of-year report, built on 245,000+ self-reported submissions and skewed toward large tech employers, shows the gradient:

Level	Median total comp	Year-over-year change
Entry-level engineer	$155K	+1.64%
Software engineer	$226K	+1.8%
Senior engineer	$312K	+4.2%
Staff engineer	$457K	+7.52%
Principal engineer	$551K	-6.58%

Across all levels, US software engineer pay rose 2.67% in 2025, per the same Levels.fyi report. Staff-level pay rose nearly three times that. The exception is the top rung: principal pay fell 6.58%, so the gradient rises through staff level rather than all the way up. SignalFire adds that top-of-band staff and principal packages at the tech majors now rival or beat director pay — a reversal of two decades of management premium.

Skills, not just seniority, carry a premium. Lightcast's analysis of over 1.3 billion job postings (July 2025) found postings requiring AI skills paid 28% more — about $18,000 a year — than postings without those skills. Lightcast's own 2026 Labor Market Predictions reports AI-skill postings grew 73% from 2023 to 2024 and another 109% from 2024 to 2025.

If your roadmap assumes a team shape you can no longer hire, our AI-powered software development teams deliver with senior engineers already working this way. Talk to us before you write another req.

Why forward deployed engineer roles grew 30% since ChatGPT launched (SignalFire, 2026)

The single most useful signal is a role that barely existed in 2019. SignalFire's 2026 data has forward deployed engineer roles up 30% since ChatGPT's launch in 2022, alongside AI/ML engineer share up 39% and research engineer roles up 28% over the same period.

An FDE sits with the customer, understands the domain, builds in the customer's environment, and owns the outcome. It is the least automatable configuration of software work: ambiguous requirements, political constraints, live feedback, and accountability for whether the thing actually worked.

That is not a coincidence. Demand concentrated exactly where the job is judgment plus context rather than translation of a spec into code. The growth of forward deployed engineering as a discipline is the clearest market evidence for the inversion thesis, because it is a role defined almost entirely by the skills AI has not absorbed.

Is AI really the cause? The honest counter-argument

Here is where most articles on this topic overreach. The claim "AI is killing junior developer jobs" is genuinely contested, and the evidence for it is weaker than the confident version you have read elsewhere.

The macro explanation came first

The tech hiring bust began before generative AI tools were in wide production use. Gergely Orosz, writing in The Pragmatic Engineer, weighed the Section 174 tax change against the macro environment as the driver of mass tech layoffs and came down on the macro side: "the end of ZIRP is more likely in my view." That is an argument about interest rates versus tax policy rather than about AI — but the sequencing is the point. The contraction was well under way before AI coding tools were in wide production use.

Indeed's own economist agrees on the sequencing. Laura Ullrich, director of economic research at Indeed's Hiring Lab, put it precisely in November 2025: "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."

Section 174 and the developer salary tax change

From July 2022, US tax rules under Section 174 forced companies to amortize software development salaries over five years for domestic staff and fifteen for foreign staff, instead of deducting them in year one. For a company whose main cost is engineers, this created tax bills on losses. Policy analyst Alex Muresianu estimated roughly 20,000 software engineering jobs lost to it.

The rule was substantially reversed in 2025, restoring immediate expensing for US-based R&D — though the fix is time-limited, with reinstatement currently scheduled for 2030. Entry-level hiring has not snapped back. That is a point for the AI explanation, but a soft one, since the macro environment stayed weak too.

The skeptics' data

Two serious research efforts find much less than the headlines claim.

Yale's Budget Lab, in its rolling analysis of Current Population Survey data through late 2025, concluded that "measures of exposure, automation, and augmentation show no sign of being related to changes in employment or unemployment." Its measure of how different the occupational mix is for college graduates aged 20 to 24 versus college graduates aged 25 to 34 has stayed in a 30-33% band since January 2021 — no visible break.

Stanford's SIEPR policy brief (July 2026) by Neale Mahoney, Erika McEntarfer, and Karsen Wahal is blunter. Since 2022, unemployment rose 0.77 percentage points for the most AI-exposed workers and 0.85 points for the least exposed. The least-exposed group did slightly worse. The brief also notes how widely adoption estimates vary by methodology: firm surveys put AI use at only about 20% of firms, concentrated in tech and finance, while a household survey found more than 40% of employed respondents using AI at work. That spread is a reminder to read any headline generative AI adoption statistic against its methodology before treating it as a measure of production use. And the brief finds measured productivity gains from AI tend to be largest for less-experienced workers — the opposite of what a pure junior-displacement story predicts.

Even the Stanford Digital Economy Lab team has narrowed its own claim. In a February 2026 update, Brynjolfsson, Chandar, and Chen conceded that with firm-time fixed effects the decline "becomes significant only in 2024; the earlier declines are likely (at least partly) due to some combination of other factors, not just AI," and stated plainly: "we do not believe that AI is always and everywhere the sole determinant of employment." They maintain the trend, now about 16% through October 2025, and report no reversal.

What we think is defensible: the recomposition is real and well measured. The attribution to AI specifically is partial, contested, and strongest from 2024 onward. Anyone selling you certainty here is selling.

Did developer employment fall, or just recompose?

The composition of engineering hiring changed far more than its volume, and that is the part the doom coverage gets wrong.

SignalFire's 2026 numbers are the clearest evidence on hiring composition:

Measure (vs 2019)	Change
Overall hiring at tech majors	Down 25%
Engineering hiring at tech majors	Down 11%
Engineering hiring at early-stage startups	Up 7%
Engineers as share of tech major hiring	55%, up from 46%
Engineering attrition rate	~9%, lowest of any function

Engineering hiring shrank less than the companies around it, grew at startups, and gained share of total hiring. Engineers are also leaving less — the lowest attrition of any function — which mechanically reduces backfill openings and makes the market feel worse to job seekers than the hiring data implies.

Be precise about what that evidence covers. Every figure above is a hiring flow, not an employment level: it tells you who is getting hired, not how many software developers are employed in 2026. The BLS figure of 1,895,500 is a 2024 base year, and the Indeed/FRED index counts postings. So the defensible claim is about composition — the mix of who gets hired has shifted decisively toward senior and AI-capable engineers — and we are not asserting a 2026 employment level, because there is no published figure we would stand behind yet.

The distribution changed too. Managers at the tech majors now supervise about 12 engineers, up from 10 in 2019, and startups average around 15, per SignalFire's 2026 report. Flatter, more senior, fewer rungs at the bottom.

Why AI raised the value of senior judgment

The mechanism is not mysterious. AI generates plausible code quickly, and verifying plausible code is a senior skill.

Stack Overflow's 2025 developer survey of 49,009 respondents found 84% using or planning to use AI tools — and only 33% trusting the accuracy of what comes out, against 46% who distrust it. The most cited frustration, at 66%, was "AI solutions that are almost right, but not quite." When developers do not trust an AI answer, 75% say they ask a person instead. On Stack Overflow's narrower "do you trust the accuracy of AI output" measure, trust fell 11 points year over year, from 40% to 29% (Stack Overflow, February 2026) — the two figures differ because the questions are framed differently, but both move the same way. The same 2025 survey data shows that distrust is highest among the most experienced developers: they are the least likely of any experience band to highly trust AI output and the most likely to highly distrust it, at more than 20%.

Google's 2025 DORA report, covering nearly 5,000 technology professionals, found 90% using AI at work and over 80% reporting productivity gains — while AI adoption remained negatively associated with software delivery stability. DORA's framing is that AI is an amplifier: strong teams get faster, weak teams get their existing problems magnified.

And a METR randomized controlled trial (July 2025) found experienced open-source developers were 19% slower on real tasks with AI tools available, while estimating they had been 20% faster. Perceived speedup and measured speedup came apart entirely.

Put those together and the demand shift makes sense. More generated code, less trust in it, unstable delivery, and unreliable self-assessment of productivity all raise the return on people who can tell good output from output that merely looks good. That capability is what "senior" means now, and it is why patterns for orchestrating AI agents in production and awareness of how models misbehave under pressure have become core engineering skills rather than research curiosities.

What to do about your specific situation

The right move depends on which constraint is actually binding. Most teams misdiagnose this.

A few positions worth stating directly, including where we are not the answer:

• If your constraint is a handful of specialist AI hires, a retained technical recruiter will probably serve you better than a consultancy. That is a search problem, not a delivery problem.
• If your constraint is that your existing seniors cannot review AI output fast enough, adding headcount makes it worse. Fix the review and testing pipeline first.
• If your constraint is that you have juniors and no ramp path, that is a training and methodology problem. A structured approach like Generative-Driven Development gives junior engineers a defined role in an AI-assisted workflow instead of competing with the model at the task it does best.
• If your constraint is delivery velocity against a fixed roadmap, embedded senior capacity is the fastest path, and that is genuinely what we do.

Do not respond to the inversion by cutting junior hiring to zero. Every organization that eliminates its entry level is borrowing against its own senior pipeline in three to five years, and the 2026 SANS/GIAC Cybersecurity Workforce Research Report, cited in CIO's coverage, already shows this happening in security, where 74% of organizations say AI is changing their team size and role structures and cuts land hardest on traditionally entry-level SOC and analyst roles.

Frequently asked questions

Is there still a software engineer shortage in 2026?

Yes, but not a general one. Entry-level candidates are abundant and senior engineers who can architect, review, and deploy AI-assisted systems are scarce. CIO's June 2026 research ranked AI/ML the hardest IT role to fill, tied at number one with cybersecurity, and dropped application development off the hardest-to-fill list entirely.

Is AI actually taking junior developer jobs?

Partly, and the evidence is contested. Stanford researchers found a 16% relative employment decline for 22-to-25-year-olds in AI-exposed occupations, but Yale's Budget Lab finds no link between AI exposure and employment change, and Stanford's own SIEPR brief found unemployment rose slightly more for the least AI-exposed workers. The end of cheap money and the 2022-2025 Section 174 tax treatment of developer salaries also explain a real share of the decline.

Should we still hire junior developers in 2026?

Yes, deliberately and with a plan. Cutting the entry level entirely removes the pipeline that produces the senior engineers who are already your scarcest resource. The change is that a junior needs a defined role in an AI-assisted workflow, plus structured review, rather than the old apprenticeship of writing the simple code a model now writes first.

How much more do AI-skilled engineers cost?

Lightcast's July 2025 analysis of 1.3 billion job postings found postings requiring AI skills paid 28% more, roughly $18,000 a year, than postings without those skills. Separately, Levels.fyi's 2025 report shows staff engineer median total compensation at $457K, up 7.52% year over year, against 1.64% growth at entry level.

Can AI close the engineering talent gap for us?

It closes part of it and widens another part. Google's 2025 DORA report found over 80% of practitioners reporting productivity gains, but also that AI adoption is negatively associated with delivery stability and mainly amplifies whatever capability a team already has. AI reduces the need for routine coding capacity and increases the need for senior review capacity.

Sizing a 2026 engineering team against a roadmap that keeps changing? Our AI strategy and roadmap work identifies which initiatives to fund first, and our GenDD training workshop gets your existing engineers working AI-natively instead of hiring around them.

Sources

• SignalFire, "SignalFire's State of Tech Talent Report — 2026", June 2026, https://www.signalfire.com/blog/signalfire-state-of-talent-report-2026
• Federal Reserve Bank of St. Louis / Indeed Hiring Lab, "Software Development Job Postings on Indeed in the United States (IHLIDXUSTPSOFTDEVE)", July 2026, https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE
• US Bureau of Labor Statistics, "Software Developers, Quality Assurance Analysts, and Testers: Occupational Outlook Handbook", August 2025, https://www.bls.gov/ooh/computer-and-information-technology/software-developers.htm
• Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, Stanford Digital Economy Lab, "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence", November 2025, https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/
• Stanford Digital Economy Lab, "Canaries, Interest Rates, and Timing: More on Recent Drivers of Employment Changes for Young Workers", February 2026, https://digitaleconomy.stanford.edu/news/canaries-interest-rates-and-timinga-more-on-recent-drivers-of-employment-changes-for-young-workers
• Neale Mahoney, Erika McEntarfer and Karsen Wahal, Stanford Institute for Economic Policy Research, "What is really happening to jobs? Separating AI hype from reality", July 2026, https://siepr.stanford.edu/publications/policy-brief/what-really-happening-jobs-separating-ai-hype-reality
• The Budget Lab at Yale, "Evaluating the Impact of AI on the Labor Market: November/December CPS Update", no publication date stated on the page; covers CPS data through late 2025, https://budgetlab.yale.edu/research/evaluating-impact-ai-labor-market-novemberdecember-cps-update
• CIO, "The 11 hardest IT roles to fill in 2026 — and what's changed", June 2026, https://www.cio.com/article/4184685/the-11-hardest-it-roles-to-fill-in-2026-and-whats-changed.html
• SANS Institute / GIAC, "2026 SANS | GIAC Cybersecurity Workforce Research Report" (source of the 60% skills-gap and 74% AI-restructuring findings cited by CIO), March 2026, https://www.sans.org/press/announcements/sans-research-cybersecurity-talent-shortage-narrative-wrong-real-crisis-what-your-team-doesnt-know-starting-ai
• Levels.fyi, "2025 End of Year Pay Report", December 2025, https://levels.fyi/2025
• Lightcast, "New Lightcast Report: AI Skills Command 28% Salary Premium as Demand Shifts Beyond Tech Industry", July 2025, https://lightcast.io/resources/blog/beyond-the-buzz-press-release-2025-07-23
• Lightcast, "2026 Labor Market Predictions", no publication date stated on the page, https://lightcast.io/resources/reports/2026-labor-market-predictions
• Stack Overflow, "2025 Developer Survey: AI", July 2025, https://survey.stackoverflow.co/2025/ai/
• Stack Overflow, "Mind the gap: Closing the AI trust gap for developers", February 2026, https://stackoverflow.blog/2026/02/18/closing-the-developer-ai-trust-gap/
• Google Cloud / DORA, "Announcing the 2025 DORA Report: State of AI-assisted Software Development", September 2025, https://cloud.google.com/blog/products/ai-machine-learning/announcing-the-2025-dora-report
• METR, "Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity", July 2025, https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
• Gergely Orosz, The Pragmatic Engineer, "The Pulse: Will US companies hire fewer engineers due to Section 174?", June 2025, https://blog.pragmaticengineer.com/section-174/
• LeadDev, "What the end of Section 174 means for software developer hiring", July 2025, https://leaddev.com/hiring/end-section-174-means-software-developer-hiring
• CIO Dive, "Tech job postings are down one-third from pre-pandemic era", November 2025, https://www.ciodive.com/news/indeed-AI-jobs-report/806111/