Software engineer job market 2026: what the data says

The software engineer job market in 2026 has more postings than a year ago, slightly fewer jobs and more announced layoffs, and it's still well below February 2020. Indeed's software development postings are about 26.5% above their May 2025 floor, but they're still 23% below February 2020. The rebound is senior-led, and the pain sits in entry-level hiring and in the number of applications chasing each role. This report is part of our Engineering management guides for CTOs and VPs of Engineering, so it's written for people who have to hire, not for people searching for a job.

The state in numbers

February 2020 is the pre-pandemic baseline because the Indeed series starts there. Where a publisher gives a 2019 figure, it appears later in the text and is labeled as 2019.

Challenger's other full years are 97,171 for 2022 and 133,988 for 2024. Its figures are announcements, not completed layoffs. The NY Fed's January 2019 reading was 4.05%. The CPS 2025 figure is an 11-month average, because no October data exists after the government shutdown.

Is the software job market recovering?

Postings say yes. Employment says not much. Both are true, and the gap between them is the main thing to watch.

Postings

Indeed's index fell from 233.84 at the 28 February 2022 peak to 61.12 on 17 May 2025, a drop of about 74%. It's since climbed to 77.32. A year ago, on 4 September 2025, it read 64.42, and on the same date this year 76.12.

Indeed's own analysis from July 2026 puts the rise at almost 15% since late February 2025, the month Claude Code launched, while postings across all occupations fell 7%. Indeed states that correlation doesn't imply causation, and so should anyone quoting that timing. At that July snapshot, software postings were 27.5% below February 2020. The September 2026 FRED reading implies 22.7%.

Employment

Payrolls tell a flatter story. BLS counts 2,355.4 thousand jobs in computer systems design and related services in September 2026 (preliminary): 5.4% above February 2020, 5.2% below the March 2023 peak, and 1.4% below September 2025. Software publishers sit 29.8% above February 2020 but have slipped about 1.7% from their November 2022 high.

CompTIA's September 2026 analysis lands in the same place from a different angle. Active IT job postings hit 625,633, a three-year high, with 47,598 new software developer and engineer postings in the month. Tech sector employment fell by 10,350 in the same month.

Postings can rise without hiring rising. Indeed measures postings, not hires, and a posting can sit open, get re-listed or never close. We can't tell from these sources how much of the gap each one explains.

Where the data contradicts the narrative

"There are no software jobs" is hard to defend with postings up about 26.5% off the floor and CPS unemployment for computer and mathematical occupations at 3.3% against 4.3% for all workers. "The boom is back" is just as hard with payrolls under their 2023 peak and postings 23% under February 2020. The accurate description is a market that stopped falling in mid-2025, has won back only about a tenth of the drop from the 2022 peak, and hasn't yet turned into net hiring.

Who is still laying off, and why?

Counts by year

Challenger counted 97,171 announced tech-sector cuts in 2022, 168,032 in 2023, 133,988 in 2024 and 154,445 in 2025. Through September 2026 the total is already 165,925, which exceeds all of 2025. Challenger's September release says that's 29% of all 2026 announced cuts and up 54% from 107,878 in the same months of 2025. Tech's share rose because tech announcements rose 54% while cuts in all other sectors combined fell about 51%.

Stated reasons

Companies give different reasons, and they don't all say AI.

Amazon's October 2025 round cut about 14,000 corporate roles. SVP Beth Galetti's memo to employees describes "reducing layers, increasing ownership, and helping reduce bureaucracy," then ties the need to be organized more leanly to AI, which it calls the biggest technology shift since the Internet.

Block put AI front and center. Jack Dorsey wrote in his post announcing the cuts that "the intelligence tools we're creating and using" make smaller, flatter teams possible, and that "we're not making this decision because we're in trouble." He described cutting from over 10,000 people to just under 6,000. Outlets including Fortune put the cut at about 40%.

Challenger tags reasons itself. AI-attributed cuts were 4,247 in 2023 and 54,836 in 2025. For January to September 2026 the figure is 120,136, about 21% of all announced cuts and the top reason so far this year. That's across all sectors, not only tech.

The AI-washing counter-view and the older explanations

OpenAI CEO Sam Altman told CNBC-TV18 at India's AI Impact Summit in February 2026 that "people are blaming AI for layoffs that they would otherwise do," while also saying some displacement is real. Martha Gimbel of Yale's Budget Lab told Fortune that "it just doesn't seem like there's major macroeconomic effects here." Both remarks appear in Fortune's report from the summit.

The same Challenger December 2025 report that counts the AI cuts also offers a second cause. Andy Challenger wrote that tech's pivot to AI, "coupled with over-hiring over the last decade created a wave of job loss." So even the AI-attribution tracker doesn't say AI alone.

Two more explanations come up. One is Section 174, the tax rule that forced companies to amortize domestic research and experimental spending instead of deducting it, starting in 2022. For tax years beginning after 2024, the 2025 budget law (section 70302) restored immediate deductions under a new section 174A; IRS Revenue Procedure 2025-28 sets out how companies make the switch. The other is interest rates. Stanford's paper tests one version of that: the young-worker gap persists when it controls for occupations' exposure to rate rises.

Can timing separate them? No. The postings low in May 2025 came after the 2022 amortization rule and before the July 2025 law restoring deductions, and a few months after Claude Code launched. That fits a tax-driven recovery, an AI-driven reshuffle toward AI work, or a plain cycle turning. Stanford's researchers add a wrinkle in the next section: the gap for young workers persists when tech firms and computer occupations are excluded, which makes a tech-only explanation harder. We can't responsibly say how much of the 2022-2025 contraction belongs to each cause, and anyone who gives a percentage is guessing.

How many software engineers are unemployed?

Nobody publishes a clean figure for software developers alone. The CPS doesn't break out a software-developer unemployment rate, and we won't offer one.

The closest official number is BLS's computer and mathematical occupations group. Unemployment there averaged 3.3% in 2025, up from 2.8% in 2024, with the number of unemployed rising from 187,000 to 228,000. All workers moved from 4.0% to 4.3% over the same two years. So the group is better off than the workforce overall, but its rate rose faster (up 0.5 points against 0.3). In September 2026 the national rate was 4.2% with payrolls up only 29,000, according to the BLS employment situation release.

The sharper numbers are for recent graduates. The NY Fed's college labor market page puts unemployment for recent graduates aged 22-27 with at least a bachelor's degree at about 5.6% in the second quarter of 2026, with underemployment at 42%. In January 2019 it was 4.05%, and in January 2020, 3.61% (NY Fed monthly series).

By major, the NY Fed's outcomes-by-major file shows this:

The file says "most recent year" without naming it, so we can't date this table with confidence. The 4.2% all-majors figure comes from the NY Fed's annual outcomes-by-major file, a different dataset from the quarterly series, and the page doesn't reconcile them.

Computer science and computer engineering graduates are more likely than most to be jobless but far less likely to be underemployed. The ones who land work earn the highest early-career wages in the table. It looks like a market that rewards the people it hires and hires fewer of them. That's a different problem from "degrees don't pay."

What is happening to junior engineers?

For the decision of whether to hire juniors, read whether a startup should still hire juniors.

Indeed's July 2026 analysis of seniority found that in software development, entry-level roles are 4.5% of postings and senior roles are 69.3% (first quarter of 2026). The mid-level share of tech postings fell 7.7 percentage points between 2019 and 2025.

SignalFire's State of Talent 2026, published 22 June 2026, puts new-grad hiring about 65% below 2019 at big tech and about 76% below at early-stage startups. Overall big-tech hiring is about 25% below 2019 on a trailing twelve-month basis, and early-stage startups are 4% below. New grads have taken a far bigger hit than hiring overall at the same companies.

The Stanford Digital Economy Lab's "Canaries" paper, revised in August 2026, is a large payroll-data study. It finds employment of 22-25-year-olds in AI-exposed occupations stands 19% below where it would be had it kept pace with less-exposed peers (ADP data through June 2026). On the same measure, using data through July 2025, the gap was 15%; earlier versions headlined a different regression estimate of 13%. The authors say the gap is larger in their ADP sample than in national surveys, and call the results descriptive, not causal. The mechanism is reduced hiring, not separations. Experienced workers show no comparable gap, and the paper finds no widespread economy-wide displacement.

The paper also lists its own limits. The effect weakens when education is controlled for. Some of the divergence started before generative AI. It persists when tech firms and computer occupations are excluded, which the authors say rules out some alternative explanations. Put together: young workers in exposed jobs are getting hired less, and the data can't yet say whether AI, a cycle or both are why.

Do companies still hire senior engineers, and who?

Yes. Across all occupations, senior is the one level where postings are growing: Indeed reports senior postings up 14.7% year over year in May 2026 and 13.5% above January 2025, while mid-level postings are 6.7% and entry-level 6.3% below January 2025. For software specifically, Indeed's earlier July analysis found that 71% of the May 2025 to May 2026 rise in postings was senior, and 37% of that rise was AI-titled. In software, which is still 23% under its 2020 level, seniors are carrying the recovery.

SignalFire's data splits by company type. Software engineers are 55% of big-tech hiring, up from 46% in 2019, even though big-tech engineering headcount is 11% lower than in 2019. Early-stage startups went the other way, with engineering headcount up 7% since 2019. Managers also oversee more people: a span of 12 at big tech, up from 10, and 15 at startups. Our read: flatter teams with more engineers per manager are one way a company can shrink and still want senior people.

We don't have sourced data on defense tech, AI-lab or non-tech enterprise hiring, so we're not going to guess which of them is hiring seniors fastest.

Why seniors? Stanford found no comparable gap for experienced workers, so employment of experienced workers hasn't weakened the way it has for the young. The data doesn't say why. Practitioners warn about the flip side: cutting the entry-level pipeline is a long-term cost. Charity Majors wrote that "by not hiring and training up junior engineers, we are cannibalizing our own future," in a June 2024 post. Addy Osmani subtitled a May 2025 essay "No juniors today means no seniors tomorrow". Those are arguments about incentives, not demand data. Indeed's figures count postings, not filled roles.

Why does a senior search feel so slow?

Partly because it is slow, and partly because the volume of applications makes it feel slower than the timelines show.

Volume

Ashby's May 2026 data release says applications per hire have tripled since 2021 and now exceed 300. Candidates are about 50% less likely to get an interview than five years ago. Kevin Connolly, Ashby's head of data, wrote that most hiring metrics "are outputs of a much larger system." The same volume that makes it harder for a candidate to be seen makes it harder for a recruiter to find the right one.

AI-generated applications and fraud

The Greenhouse AI in Hiring report (4,136 respondents in the US, UK, Ireland and Germany, published 19 November 2025) found 91% of recruiters had seen candidate deception. Of recruiters, 34% spend up to half their week filtering spam, and 41% of US job seekers said they'd used prompt injection in applications. That's survey data from one vendor, and both sides of the market have reasons to report the other's bad behavior. Even so, the direction is plain: screening takes longer when many applications are machine-written or fake. We cover detection in how to catch fake candidates.

Time to hire

Ashby's figures are 8 weeks to first fill for business roles and 10 weeks for technical ones, and technical roles need nearly twice the interview time. Ashby also says hiring timelines have stabilized. That contradicts the common claim that processes keep getting longer: on Ashby's data they're long but have settled at those levels. We found no source for a rising number of interview rounds, so we aren't claiming one.

Ghost jobs

Greenhouse's survey found that 69% of US job seekers had met fake job postings. That's a perception of scams, which overlap with ghost jobs but aren't the same thing. We found no primary dataset for what share of postings are ghost jobs, so the share is unknown.

How long a search takes

No primary dataset gives the duration of an engineer's job search. The only official figure is BLS's average duration of unemployment, 24.8 weeks in September 2026 (median 11.5 weeks), and that covers all workers, not engineers. Vendor blog numbers for engineer search length don't have sources behind them, so we're not repeating them.

Practitioners describe it as harder than the data does. A 15-year senior engineer wrote a Hacker News post in August 2025 about applying to roughly 450 positions before being hired. Greenhouse CEO Daniel Chait told Fortune, in a July 2026 piece, that "the market just isn't working for either side." Those are single accounts, but they match the volume data above.

What the executives and practitioners say

The pessimistic case

Dario Amodei said at the Council on Foreign Relations in March 2025: "we'll be there in three to six months" in a world "where AI is writing 90 percent of the code." The CFR event page hosts the recording. The window closed in September 2025. We have no industry-wide measure of AI's share of code to check it against, so we can't say whether it came true. Employment data doesn't test his claim either way. Anthropic sells AI coding tools, including Claude Code, so he has a stake in the forecast.

Mark Zuckerberg said on The Joe Rogan Experience on 10 January 2025 that "an AI that can effectively be a sort of mid-level engineer" would exist at Meta and other companies that year, as ITPro reported. The deadline's passed. Indeed shows mid-level postings across all occupations 6.7% below January 2025 and, across tech postings, their share down 7.7 points since 2019, which is consistent with weaker mid-level demand but doesn't show the cause. We didn't find a 2026 Zuckerberg statement confirming that the prediction came true. Meta builds AI models, so he has a stake here too.

The optimistic case

Jensen Huang said at NVIDIA's GTC Taipei keynote on 1 June 2026: "It's causing more software engineers to be hired." He called talk of AI reducing jobs "complete nonsense," according to a third-party transcript of the keynote. NVIDIA sells the hardware that AI coding runs on, so he benefits if the market believes that. The data does show senior postings up, but the BLS payroll series doesn't show a surge in software engineers.

Sundar Pichai made a narrower version of the same point to Lex Fridman in mid-2025. Google had seen a 10% engineering velocity increase from AI, he said, "but we plan to hire more engineers next year," because "the opportunity space of what we can do is expanding too," according to the interview transcript. Google sells AI too, so the same caveat applies, though his claim is about Google's own hiring rather than the whole market.

The over-hiring and pipeline case

Andy Challenger and Sam Altman, quoted earlier, both give partial credit to causes other than AI. AWS CEO Matt Garman went furthest on the junior pipeline. On Matthew Berman's show in August 2025, as The Register reported, he called replacing juniors with AI "the dumbest thing I've ever heard" and asked "How's that going to work when ten years in the future you have no one that has learned anything". His company also cut about 14,000 corporate roles two months later, for stated reasons that mixed layers and bureaucracy with the pace AI sets. Those two facts don't contradict each other, but they show how little a single quote proves.

Garman, Majors and Osmani warn about the pipeline, and none of the executives quoted disputes it. On how many jobs AI is cutting today, they disagree, and the data here doesn't settle it.

What this means for your hiring plan

Layoffs don't make senior engineers easy to hire. The inbound pool is large (Ashby reports over 300 applications per hire), but Indeed's data shows employers competing for seniors, with senior postings up 14.7% across all occupations while mid-level and entry-level postings sit below January 2025.

As judgment, not fact: budget ten weeks or more for a senior technical hire, since Ashby's figure is for first fill and the senior tail is likely longer. Run an engineer-led screen and a referral channel alongside your inbound, because a large pool isn't a large qualified pool. Treat the junior pipeline as a separate plan with its own budget and owner, since SignalFire's data shows new-grad hiring has fallen further than hiring overall and no one is replacing it by default.

Before you rewrite the senior screen, read what to test when you hire a senior now, which covers how strong seniors work with coding agents and what that looks like in an interview.

FAQ

Is the software engineering job market recovering in 2026?

Partly. Indeed's software development postings index was 77.32 on 18 September 2026, up from a 61.12 trough in May 2025 but 23% below February 2020. BLS payrolls for computer systems design are 1.4% below a year ago. Postings are recovering faster than employment, and Indeed's postings measure demand, not hires.

What is the unemployment rate for software developers?

No official rate exists for software developers alone, because the CPS doesn't publish one. The nearest BLS figure is computer and mathematical occupations, at 3.3% in 2025 against 4.3% for all workers. The NY Fed reports 7.0% for recent computer science graduates and 7.8% for computer engineering, from its most recent annual data.

Are companies still hiring senior software engineers?

Yes. Indeed's July 2026 analyses found senior postings up 14.7% year over year across all occupations in May 2026, while mid-level postings were 6.7% and entry-level 6.3% below January 2025. About 71% of the rise in software postings over May 2025 to May 2026 was senior. SignalFire adds that software engineers are 55% of big-tech hiring, up from 46% in 2019.

How long does it take to hire a senior software engineer?

Ashby's May 2026 data shows about 10 weeks to first fill for technical roles, and 8 weeks for business roles. It says hiring timelines have stabilized. No primary dataset gives a senior-specific figure, so our planning assumption is 10 weeks as a floor. Candidate volume of over 300 applications per hire adds screening work, not speed.