Here’s the uncomfortable truth about every “AI will take your job” headline: most of them are written by people who haven’t read the actual reports.
The Bureau of Labor Statistics, the World Economic Forum, and McKinsey have all published their 2025–2026 data. I’ve gone through it. And the picture is a lot more interesting — and a lot more specific — than the panic merchants would have you believe.
This isn’t a feel-good piece about how “developers will always be needed.” That kind of vague reassurance is useless if you’re trying to make a real decision about your career. This is a breakdown, field by field, of what the hard numbers actually say.
Let’s get into it.
First, the only number that matters from the WEF
The World Economic Forum’s data, updated at Davos 2026, projects that AI will eliminate 92 million jobs by 2030 while creating 170 million new ones. That’s a net gain of 78 million positions across the global economy.
But here’s what the headlines always leave out: those 170 million new roles are not evenly distributed. They cluster hard around specific fields. And tech — specifically AI, data, and cybersecurity — accounts for a disproportionate share of them.
The question isn’t “will tech survive?” It will. The question is which corners of tech are structurally protected and which are quietly getting hollowed out.
The short-term picture (right now through 2028)
These are the fields where the data shows strong, current, immediate demand — and where structural barriers make AI replacement genuinely difficult in the near term.
1. Cybersecurity — and the numbers are not close
The BLS 2024–2034 projections put information security analysts at +29% growth. That is the single fastest-growing occupation in the entire computer and IT category. Not fast — fastest.
Why can’t AI just do this? Because cybersecurity is adversarial by nature. The moment an AI defense becomes widely deployed, attackers study it and adapt. It’s not a problem with a stable answer that can be automated away. It’s a permanent arms race, and human judgment at the edges of that race is irreplaceable.
What’s also happening is a split inside cybersecurity itself. The jobs aren’t just growing — they’re differentiating into higher-value lanes: AI red team specialists, AI security engineers, cloud exposure analysts, evidence-driven validation practitioners. The field is getting richer, not thinner.
2. Embedded systems and firmware engineering
There’s no BLS headline number for this one, but the structural logic is solid. Embedded systems work — the code that runs inside medical devices, industrial machinery, automotive systems, and IoT hardware — operates under constraints that current AI tooling simply cannot navigate.
Real-time requirements. Proprietary toolchains. Physical hardware integration. Safety certification processes that require human accountability. These aren’t soft barriers. They’re hard walls that won’t come down soon.
The demand driver is obvious: the IoT market, robotics, and autonomous systems are all expanding, and every one of those systems needs embedded engineers to build and maintain the software stack underneath.
3. ML/AI engineering — the self-reinforcing demand
Every major tech company — Meta, Google, Microsoft, Amazon — reported in 2026 that ML engineering is their top hiring priority. This one almost doesn’t need explaining. AI is building AI. The people who know how to design, train, evaluate, and deploy machine learning systems are the people steering the machine.
The irony is real: AI is displacing some roles while simultaneously creating enormous demand for the people who build it.
4. Data engineering
This is the unsexy underbelly of the AI boom, and it’s one of the most protected fields in tech right now.
AI systems are only as good as the data pipelines feeding them. Data governance, data quality, pipeline architecture, and infrastructure reliability are not problems AI solves — they’re problems AI creates more of. Every new model deployment generates new data engineering work.
The BLS specifically called out data scientists as the fourth fastest-growing occupation in the entire US economy. Data engineers sit directly below that layer, and the demand is just as real.
5. Computer and information research scientists — +20% projected
This one is straightforward. Foundational research into new algorithms, new computing paradigms, new approaches to machine learning — this is work that requires genuine novelty. It cannot be automated because the task is, by definition, to produce something that doesn’t yet exist.
The BLS projects 20% growth for this category through 2034. That’s more than six times the average rate of growth for all occupations.
The long-term picture (2028 and beyond)
Some fields are not just safe in the short term — they have structural reasons to stay protected for a decade or more.
Cybersecurity (AI security is a new sub-field, not a threat)
The emergence of AI as a tool for both attack and defense doesn’t shrink the cybersecurity field. It expands it. AI red-teaming, adversarial machine learning, and AI model security are genuinely new specializations that didn’t exist five years ago and now command serious salaries. The +29% projection already accounts for some of this, but the real growth in these sub-fields likely outpaces even that number.
Robotics and autonomous systems engineering
To build a functional autonomous system, you need someone who understands mechanical engineering, electrical engineering, embedded software, sensor fusion, and safety validation — often simultaneously. AI can assist with parts of this. It cannot own the whole stack. The integration work, the physical debugging, the safety certification — these require people who can move between domains in ways that current AI cannot.
Healthcare IT and medical device software
FDA regulatory constraints create a compliance moat that will protect this field for years. Every software change in a Class II or Class III medical device requires documented verification and validation processes with human sign-off. The liability structure of healthcare IT means the industry will be slow to automate away human engineers regardless of what the technology can theoretically do.
HCI and UX engineering
User research, cognitive load analysis, accessibility work, and the translation of messy human behavior into software design decisions — this requires understanding humans, not just code. AI can generate UI layouts. It cannot tell you why users are abandoning a checkout flow at step three after the last redesign, and it cannot run the discovery process that surfaces that insight.
The one category people miss: “AI-accelerated, not displaced”
There’s a third bucket that doesn’t get enough attention. These are the roles where AI increases productivity rather than reducing headcount — which means the people in these roles become more valuable, not less.
The BLS projections note explicitly: “Some computer occupations whose skills will be increasingly needed to meet the growing demand for AI-based systems… are expected to experience a positive impact to their employment outlook.”
Specifically:
- Senior software developers who can direct AI tooling toward correct solutions are more valuable than ever, because the firms that deploy AI effectively need fewer junior headcount and more senior judgment
- AI product managers who have absorbed the workflow work that junior PMs used to handle and now operate at a higher strategic level
- Cloud infrastructure engineers who keep AI workloads running reliably at scale
The part nobody wants to hear
The BLS and McKinsey data both point to the same uncomfortable split.
Senior and specialized roles across all of the categories above are stable or growing. The demand is real, the projections are strong, and the structural barriers are genuine.
Entry-level roles are a different story. Indeed’s Hiring Lab data from January 2026 shows roughly 13–28% fewer entry-level tech postings compared to the 2022 peak. McKinsey puts it plainly: 30% of the task composition for very junior workers is “ripe for automation.”
This doesn’t mean entry-level is dead. It means the path in is narrower and requires deliberate positioning. The developers getting hired at junior level in 2026 are the ones who have already picked a lane — security, embedded, data, ML — and are not presenting as generalist coders.
The bottom line
If you’re making career decisions based on vibes and LinkedIn panic threads, stop. The official data is public and it’s specific.
The fields with the strongest structural protection:
- Cybersecurity (+29% BLS, adversarial nature, AI security as a growth sub-field)
- Computer and information research science (+20% BLS, novelty-dependent work)
- ML/AI engineering (top hiring priority at every major tech company in 2026)
- Data science / data engineering (#4 fastest-growing in the US economy)
- Embedded systems / robotics (physical integration barriers, IoT expansion)
- Healthcare IT (regulatory moat, liability structure)
The honest caveat: specialization is no longer optional. The broad “I know how to code” identity is under pressure. A sharp focus on one of the fields above is not just career advice — it’s what the data actually shows separates the roles that are growing from the ones that are shrinking.
The AI revolution is real. So is the job market for the people who know how to work with it, build it, secure it, and keep it running.
Data sources: BLS Occupational Outlook Handbook, BLS Employment Projections 2024–2034, WEF Four Futures for Jobs 2030, WEF Davos 2026, Indeed Hiring Lab Jan 2026, McKinsey State of AI 2025