The Tech Job Market Didn't Collapse — It Split in Two. Here's Where Career Changers Should Aim

Here’s the paradox that should reframe how every career changer reads the news. In the first four months of 2026, the tech industry shed something like 78,000 jobs, a large share of them attributed to AI and automation. Over the same stretch, hundreds of thousands of AI-related roles sat open and postings for them kept climbing. Mass layoffs and a hiring boom. Same industry, same moment, both true.

If you only read the layoff headlines, you conclude it’s too late to get into tech. If you only read the AI-boom headlines, you conclude you should quit your job tomorrow. Both conclusions are wrong, because they treat “tech jobs” as one market. It isn’t one market anymore. It split in two, and knowing which half you’re aiming at is the most important strategic decision a career changer makes.

The two-economy thesis, with data

The split is visible in the numbers once you stop averaging them together:

• The generalist-junior lane is contracting. Software-developer employment for the 22–25 cohort fell roughly 20% since 2024 (Stanford HAI’s 2026 AI Index). General software postings sit well below pre-pandemic levels. The entry-level “write standard code” job is genuinely harder to land than it was.
• The AI-adjacent lane is growing. LinkedIn data from early 2026 showed AI/ML engineering postings up around 34% year over year while overall tech postings fell about 8%. CBRE’s tech-talent work tracked AI-related postings climbing from about 11% of US tech ads in 2022 to roughly 20% by mid-2025, even as the supply of AI-skilled workers grew more than 50% year over year.

Average those two lanes together and you get a muddy “tech is flat” story that’s useless for decisions. Pull them apart and the strategy is obvious: don’t aim at the lane that’s shrinking.

”Is data engineering dying?” — no, it moved up the stack

A specific anxiety worth addressing, because people keep googling it. AI can now generate a lot of standard pipeline and glue code, so it’s tempting to conclude data engineering is doomed. The reality is the opposite of dramatic: the commodity part of the role — hand-writing routine ETL scripts — is being automated, which pushes the human work up the stack toward systems design, data architecture, reliability, and governance. The role isn’t dying. It’s getting more senior in its center of gravity, and the low-value part is evaporating. That’s a pattern, not an exception — it’s happening across the field.

What this means tactically mid-pivot

If you’re partway into a career change right now, the split changes your targeting:

• Stop optimizing for the contracting lane. Don’t model your plan on “become a generic junior developer.” That’s the role under the most pressure.
• Aim at AI-adjacent and AI-augmented work — the roles where the postings are actually growing, including the whole middle tier that isn’t pure ML engineering.
• Lead with what compounds. The roles that are growing reward judgment and domain expertise layered with AI fluency — not raw coding speed, which is exactly what AI commoditized.

The amplification framing

Here’s the mental model that ties it together: AI rewards people who already have domain judgment. It’s an amplifier. Point it at someone with real experience in a field — marketing, finance, operations, law — and it multiplies their output. Point it at someone with no judgment to amplify and it just produces confident nonsense faster.

That’s genuinely good news for a career changer with a prior career. Your old domain isn’t dead weight; it’s the thing AI amplifies. The skills I carried out of ad tech are worth more in an AI-augmented role than they were on their own. The losers in this split are people whose only asset was being a cheap pair of hands for routine code. The winners are people with judgment plus AI fluency.

A positioning checklist

To put yourself on the right side of the split:

• Target AI-adjacent or AI-augmented roles, not generic-junior ones.
• Make your prior domain expertise explicit — it’s the thing being amplified.
• Build demonstrable AI fluency (use and failure modes), not just enthusiasm.
• Show judgment with proof of work, not credentials alone.
• Frame yourself as the intersection — domain + AI — because that intersection is scarce.

The collapse story and the boom story are both real. The only mistake is treating them as one market and aiming at the wrong half. Aim at the half that’s growing, bring the judgment you already have, and the split stops being a threat and starts being the opening.

---

This is the competitive pillar of an ongoing, in-public career change. Start at the cornerstone plan, or read about the roles that reward a non-technical background.