Software development job postings in the US are up almost 15% since Claude Code launched in February 2025, even as postings overall fell 7% over the same stretch, according to Indeed's Hiring Lab. Read as a headline, that sounds like relief: the market that spent 2025 leaving reqs open for months is finally loosening. It isn't, not for the person actually doing the hiring. Indeed's own breakdown of where that growth landed shows it concentrated almost entirely in two overlapping groups: senior roles, and roles with the word "AI" in the title. If the seat you're trying to fill is neither, the recovery happened somewhere else. If it's both, you're not hiring into an easier market. You're hiring into a smaller one with more people standing in it.
The recovery is real. The catch is in who it counts for.
Indeed indexes job postings against a baseline of February 1, 2020, before the pandemic reshaped hiring for a generation. On that index, software development postings sat around 73 as of June 2026, which is to say still roughly 27 to 30% below where they started. Overall postings, by contrast, are essentially back to the same level as February 2020, at 101 on Indeed's index (up 1% for the month, down 3.7% year over year). Software development is one of the few categories still climbing out of a hole, while the broader engineering and healthcare fields Indeed tracks separately are running about 30% above that same baseline.
So "recovering" and "recovered" are doing different work in that sentence. Software dev postings are climbing out of a deep trough. They have not climbed back to their old level, and the climb has been narrow rather than broad, which is the part that actually matters if you're the one writing the req.
Two numbers that explain nearly the whole story
Here's the detail that turns a vague trend into an operating fact. Of the increase in US software development postings between May 2025 and May 2026, 71% came from senior-level roles. Separately, 37% of that same increase came from postings that mention AI in the title. Those two figures aren't additive, since a senior role hiring for production AI work counts in both columns at once, and that overlap is exactly the point: the growth isn't spread across the profession, it's stacked on a narrow intersection of seniority and AI fluency.
The same pattern shows up outside software specifically. Across all occupations Indeed tracks, postings mentioning AI hit 5.9% of the total in June 2026, more than double the previous peak of 3.3% set back in 2022. AI-adjacent hiring isn't a niche inside a niche anymore. It's the fastest-growing part of a labor market that is, everywhere else, close to flat.
Why concentration is bad news for the hiring manager, not good news
If you run a Series A product company and you've been telling your board that hiring should get easier once the market "normalizes," this is the data that says otherwise. Every other funded AI-native company chasing the same roadmap is fishing in the identical, shrinking pond: senior engineers who've actually shipped a production AI system, not toured one in a hackathon. Demand didn't spread out as the market recovered. It concentrated.
It's also worth saying plainly that the layoff headlines don't rescue you here either. Overall postings are down year over year, not up, which means the market isn't quietly flooding with laid-off senior talent looking for a landing spot. The honest read is that 2026 is a harder year to hire a senior AI engineer than 2025 was, not an easier one, because everyone with funding and a roadmap is competing for the same narrow slice at the same time.
Eleven weeks, one small team, and a board meeting on Friday
Picture a 24-person Series A company building evaluation tooling for teams shipping LLM features into production. Its roadmap needs two senior engineers who've actually run inference at scale, not two more generalists. The VP of Engineering opened both reqs eleven weeks ago. One offer went out in week seven and was declined for a counter from the candidate's current employer. The other candidate is still mid-loop, because the fourth interviewer keeps rescheduling around a launch.
Nothing about that process was badly run. The reqs were well-scoped, the comp was competitive, and the interview loop was tight. The problem sits upstream of all of it: there are simply fewer people who match "senior" and "has shipped production AI" than there are funded companies who need exactly that person this quarter, and no amount of process discipline manufactures more of them.
What actually shortens the wait, and where it doesn't
The fix isn't a better job posting. It's decoupling the deadline from the search. A vetted staff-augmentation bench can put a senior engineer with real production AI experience into your standup inside days to two weeks, which buys the search room to be patient instead of desperate, the same logic behind what an open senior role actually costs. Getting that arrangement to actually work in week one matters as much as the placement itself, which is what a properly run first week is for.
Here's the honest exception, because a pitch that fits every situation is selling, not advising: if the seat you're trying to fill will set the architecture your company runs on for the next three years, a bridge doesn't solve your problem. It just delays deciding who that person actually is, and that decision still has to be yours. Augmentation keeps the roadmap moving while you find that person. It isn't a substitute for finding them, and the trade-offs between that kind of standing arrangement and a fixed-scope build are worth reading before you pick one, in fixed-scope versus dedicated teams.
None of this means the senior AI engineering market is unsolvable, only that it's smaller and more contested than the recovery headlines suggest. If you've got a req that's been open long enough to show up in a board deck, our Silicon Valley team can tell you, honestly, whether a bridge makes sense or whether the search just needs to run longer. Tell us what the seat needs to ship and we'll give you a straight read within 48 hours.
Sources
- Indeed Hiring Lab: AI and job postings: from destruction to creation?
- Indeed Hiring Lab: US labor market snapshot, June 2026