The single most-predicted casualty of artificial intelligence — the software developer — is quietly becoming the labour market’s comeback story. The question is for whom.
CNBC’s 11 October analysis of fresh Indeed Hiring Lab and Bureau of Labor Statistics data found US software-development job postings have risen nearly 15 percent since February 2025 — the month Anthropic’s Claude Code launched — while overall job postings fell 7 percent over the same period. Behind that headline sits a split this site has tracked for months: the same datasets showing the rebound is happening fastest in the profession where AI writes the most code.
🔍 THE BOTTOM LINE
🔍 THE BOTTOM LINE: U.S. software-development postings are up almost 15 percent since February 2025 against a 7 percent decline for jobs overall, but Indeed’s decomposition shows 71 percent of the year-to-year gain came from senior roles and 37 percent from jobs with AI in the title — while Stanford’s index has employment for 22 to 25-year-old developers down about 20 percent. The profession is pricing its experienced practitioners up, and pricing its beginners out.
The rebound nobody forecast
Hiring Lab’s own July 2026 analysis, “AI and Job Postings: From Destruction to Creation?”, put numbers on a change that had been building for a year: over the past 12 months the occupations most exposed to AI saw the largest rebounds in postings — inverting a 2022-2026 relationship in which more exposure meant steeper declines. Software development is the sharpest example, and Indeed is careful with the causality: Claude Code’s launch and the rebound landing in the same month is “a coincidence that cannot be ignored”, but not a proven link. The analysis also finds the pattern is not unique to the US — the share of postings that are software jobs has been rising in most large developed economies, with English-speaking countries trending more consistently positive.
The rebound is real but partial. Even after the rise, software-development postings remain about 27.5 percent below their pre-pandemic level, from a low start point; McKinsey Global Institute’s research, cited by CNBC, notes the earlier decline was not purely an AI story — it came off a 2022 pandemic-hiring peak when tech companies pulled back en masse.
The best one-line summary came from McKinsey’s Eric Kutcher on CNBC’s “Squawk on the Street”: “We’re seeing much more growth in the places we thought we were gonna see the biggest displacement.”
Why more automation has meant more hiring
The mechanism, per Indeed economist Sneha Puri and the researchers CNBC assembled, is that AI is changing what firms need rather than whether they need people. Companies are producing more code than ever; someone must evaluate AI-generated output, connect tools to wider systems, and help businesses apply them. Demand has concentrated on exactly those tasks.
The automation data itself supports the shift. GitHub reported on 7 October that one in three pull requests on the platform now involves an AI agent, up from fewer than one in 10 a year earlier — with the platform warning that “much of it may never be fully read by a human” if that pace holds. That has created a widening pool of oversight work: reviewers, security specialists, integrators. Anthropic’s own research finds humans are still needed to supervise AI-written code — while cautioning it is “by no means certain that this pattern will persist into the future.”
The pay data points the same direction. Indeed Hiring Lab’s September analysis found advertised pay in the most AI-exposed US occupations has climbed 46 percent since 2021 against 25 percent in the least-exposed — a 21-point premium that includes software development among the fastest risers (/indeed-ai-exposed-pay-premium-2026/).
The beginners’ squeeze inside the boom
Here is the part of the story with teeth. Stanford University’s 2026 AI Index, cited by CNBC, finds employment in AI-exposed fields for workers aged 22 to 25 has fallen 16 percent since generative AI arrived — and early-career software developers specifically have seen a steeper 20 percent decline even as senior developer roles expand. The door did not close; it stopped opening (/stanford-canaries-ai-entry-level-gap-widens-19-percent-2026/).
Indeed’s own entry-level share data, reported here in September, showed salaried entry-level postings in AI-exposed occupations falling from 29 percent of the total in 2021 to 10 percent by 2026 (/coding-not-solved-junior-careers-2026/). Puri’s advice to students reflects both sides at once: build familiarity with AI tools, but do not neglect “communication, interpersonal skills and leadership” — and her warning to employers is blunter: “Today’s entry-level talent becomes tomorrow’s senior experts. If employers don’t build those pipelines now, they’re not going to have senior talent they need 10 years down the line.”
That is the same two-margin pattern the CEPR/VoxEU task-level analysis showed across the whole US economy in early October: AI-substitutable work shrinking 3.3 points a year while AI-complemented work grows just as fast, with the adjustment landing on hiring rather than firing (/ai-two-margins-hiring-not-firing-youth-2026/).
The long-run forecast says growth — more than twice the national pace
The official record, for balance: the US Bureau of Labor Statistics’ August 2026 update projects total US employment will grow 3.5 percent from 2025 to 2035, with computer and mathematical occupations — software developers among them — the fifth-fastest-growing group at 7.3 percent, a pace the release directly attributes to “the continued proliferation of digital tools and AI solutions across the economy.” A profession growing at more than double the all-occupation average while AI writes a rising share of the code is not the headline anyone would have drafted in 2023.
So the story of 2026 is not “AI killed the developer” nor “nothing happened.” It is that AI has split one career into two labour markets: a senior tier riding a posting-and-pay boom, and an entry tier carrying nearly all the adjustment. That split is now showing up in other white-collar occupations too — Indeed notes AI mentions in job titles are spreading well beyond tech.
What it means for New Zealand job seekers
The NZ angle is a smaller and skewed version of the same numbers. SEEK’s August report showed NZ job ads up 9.7 percent year on year with AI skills now in 4.0 percent of listings, up 93.3 percent a year (/nz-seek-job-ads-august-ai-skills-2026/) — concentrated in ICT, where 17.6 percent of ads mention AI. Hays’ FY26/27 salary guide found 82 percent of NZ employers reporting tech capability gaps, meaning the senior squeeze is even tighter here than the US data suggests, while only 22 percent of NZ tech professionals have had formal AI training. The New Zealand software market is small enough that both edges of the American pattern show up at once: hard-to-hire seniors and a thinning junior ladder.
For a Kiwi developer or IT graduate reading this, the Indeed and Stanford data combine into one practical read — the demand is real, but it is priced for people who can already verify what the machine wrote. A graduate whose GitHub shows working AI-assisted projects is applying to the 37 percent of the rebound that mentions AI in the title; a CV with only coursework is applying to the tier that is disappearing.
❓ FAQ
Are software developer jobs growing or shrinking in 2026? Both, by cohort. US software-development postings rose nearly 15 percent between February 2025 and mid-2026 while overall postings fell 7 percent, and BLS projects computer and mathematical occupations growing 7.3 percent from 2025 to 2035 — more than twice the all-occupation pace — but the near-term gains have gone mostly to senior and AI-titled roles, with employment for 22 to 25-year-old developers down about 20 percent (cnbc.com).
Is AI writing most of the code now? On GitHub, one in three pull requests now involves an AI agent, up from under one in 10 a year ago; some engineers claim AI-written code passed the halfway mark, which GitHub has not confirmed. Its own statement warns most pushed code could be agent-written within two years if the pace holds.
Why are senior developers in demand if AI can code? Because the scarce skill has shifted up the stack: evaluating AI output, setting architecture, and integrating tools into enterprise systems — work requires the judgement that experience builds. Hence 71 percent of the posting rebound coming from senior roles.
What should a New Zealand tech graduate do with this? Target roles where AI is part of the job title rather than the threat — NZ SEEK data shows AI mentions across 4.0 percent of ads, doubled annually — and get projects demonstrably built with AI tools in front of employers, since the entry tier that trains beginners inside companies has thinned from 29 percent to 10 percent of postings in exposed occupations.
Does BLS still think software development is a growth career? Yes — its August 2026 projections put computer and mathematical occupations (software developers included) among the fastest-growing major groups at 7.3 percent from 2025 to 2035, more than double the 3.5 percent all-occupation pace.
🔍 THE BOTTOM LINE
The developer-rebound data is the strongest “jobs created, not destroyed” evidence the AI transition has produced so far — and the junior collapse inside it is the strongest evidence that AI’s costs are landing precisely, not evenly. The two halves settle the same practical question for readers in NZ and abroad, and the honest 2027 test is whether the senior boom eventually re-opens the junior door or whether oversight work itself gets automated before the pipeline refills. Watch the entry-level share of postings, not the headline: if it stays pinned near 10 percent while the profession grows, the split has hardened into structure.
📰 Sources
- CNBC — AI is automating coding. So why are software developer job postings on the rise? (11 October 2026)
- Indeed Hiring Lab — AI and Job Postings: From Destruction to Creation? (8 July 2026)
- Indeed Hiring Lab — AI Exposure Isn’t Squeezing Advertised Pay in the US (17 September 2026)
- GitHub — Secret protection must scale with software (7 October 2026)
- Stanford HAI — 2026 AI Index, economy chapter
- US Bureau of Labor Statistics — Employment Projections, August 2026 — CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
