The 10 Highest-Paying Growth Jobs and 5 Fading Occupations: What BLS Projections Through 2035 Show

The video opens with a single tension: more than 750,000 office and administrative support jobs are expected to vanish from the US economy by 2035, while a handful of occupations are projected to expand by over 40 percent. The host frames the piece as a decision guide for students, career switchers and anyone wondering which skills will still pay in five years; the method is explicit — ten high-paying growth careers ranked by median salary, followed by the five fastest-shrinking occupations and a simple exercise to rethink your own tasks through a new lens.

Why high AI exposure does not kill growth

The methodology is built on Bureau numbers: the office scores 831 occupations for how much of the work inside each job could be done or assisted by AI, drawing on five research sources including cloud-usage data and Copilot task mapping. The surprise is quantitative — among the 206 occupations with the highest exposure, 141 are still projected to grow over the next decade, about 68 percent. The host stresses the distinction that matters: augmentation of tasks inside a job is not the same as replacement risk, and high exposure is not an automatic disappearance signal.

The bottom of the top ten breaks the usual tropes. Tenth is management analysts — consultants by another name — with median pay a touch above 100,000 dollars; despite very high exposure, because deliverables are research, analysis and slides, the outlook is about 109,000 added jobs and roughly 94,000 openings a year. The explanation is straightforward: almost every company is trying to reorganize around AI and few know how, so the reorganization itself becomes the product — a line echoed by a major tech CEO pushing back on white-collar apocalypse framing in June. Ninth is the bet most lists skip, health educators who train nurses, pharmacists, physical therapists and physician assistants, projected near 17.9 percent growth; health care is adding roughly a third of all new US jobs and clinical training legally requires supervised human hours that cannot be scaled with a chatbot — the barrier is real, a doctoral or professional degree, making this a decade-long play, not a six-month switch.

Eighth leaves the laptop economy entirely for the job site; construction managers who handle budgets, schedules, subcontractors and inspections are projected to add about 55,000 jobs with near 50,000 openings a year. The through-line is AI's physical footprint — data centers, power plants and chip fabs turn every AI announcement into a construction project with a manager in a hard hat — a point underscored by a chipmaker's late-August remark that chip, packaging and AI factories are already creating hundreds of thousands of roles. The money follows the same direction; a telecom giant is reported to be earmarked roughly 38 billion dollars spread over five years to recruit and skill blue-collar and trade workers, and the host notes that a generation told the safe path was a screen job may find the next decade arguing the opposite.

The two ends of data and the backbone of health

Seventh is data scientists, the third-fastest-growing occupation in the country at about 34.6 percent, from roughly 275,000 to 371,000 by 2035, and the immediate question is how that can be true when AI writes code and runs analyses. The answer is that automation ate the middle — cleaning, boilerplate, first-pass modeling — but left the two ends: deciding which question is worth asking and judging whether the answer is trustworthy enough to bet the company on. An economist on the podcast frames the mechanism as an economic complement: when the 30 to 40 percent of work that AI can cheapen gets cheaper, the remaining 60 percent that humans do becomes more valuable, not less — those still coding as if it were 2022 are at risk, while those who let AI handle the automatable slice and build skills for the rest pull ahead.

Sixth is a non-clinical route into health care that needs a bachelor's, not a medical degree; medical and health services managers who run clinics, hospital departments, nursing facilities and group practices are projected to grow about 24.2 percent with 155,000 new jobs and 62,000 openings a year. This is where two unstoppable forces meet: aging-driven demand exploding on one side and the most regulated, fragmented and administratively broken industry being digitized badly and expensively on the other — whoever bridges the clinical and the operational becomes the bottleneck, and bottlenecks are paid well. Fifth is information security analysts at about 21 percent growth; the standard story of more hackers is true but incomplete, because AI made attacking cheap — flawless English phishing at scale, voice cloning, automated vuln scanning — and firms stuffing AI into everything create novel risk categories with no playbook yet. A caveat is added: at about 193,000 workers and 14,000 openings a year the field is smaller and competitive, yet often flagged as the most accessible high-paying path on the list without a heavy degree requirement.

Clinic and code: why health and software lead growth

Fourth is nurse practitioners at about 132,000 dollars a year and 29,000 openings, requiring a master's and carrying high exposure yet posting about 41 percent growth — the fastest rate in the country and the clearest proof that exposure and risk diverge. The drivers are a physician shortage, expanding state scope-of-practice laws and an aging population needing chronic-care management beyond capacity; AI can draft notes and flag interactions but it cannot perform an in-person exam, assume legal responsibility for a diagnosis, or stay with a scared patient for twenty minutes — a diagnosis is just one duty within the full role, not the role itself. Third stays in health care with physician assistants at about 136,000 dollars and roughly 20 percent growth with 35,000 added jobs, sitting in the bottom quartile of exposure among 831 occupations — based partly on what people actually ask assistants to do — making it one of the rare combinations of high pay, solid growth and low touchability by current AI.

Second is the most obituarized role, software developers at about 135,000 dollars, 10 percent growth and 95,000 openings a year for a total of about 174,000 added jobs — the largest absolute gain among occupations paying over 100,000, lifting headcount from roughly 1.7 million to 1.9 million by 2035. The picture underneath is mixed: one venture dataset finds big-tech hiring down about 25 percent since 2019 while engineering hiring is down only 11 percent and engineers rose from 46 to 55 percent of new hires — the most resilient function, not the most fragile. Another hiring lab finds software postings up about 15 percent since a leading coding agent launched while overall postings fell 7 percent, but 71 percent of that gain is concentrated in senior roles — the job is growing while the entry point narrows.

The top: why deciding beats doing

First is computer and information systems managers at about 175,000 dollars, 15.8 percent growth and 108,000 new jobs — the person who decides what a company builds, buys, retires and what happens when it breaks, from IT directors to engineering managers and midsize-firm CTOs, rated very high exposure for its document, planning and review load yet still growing nearly 16 percent at the top of the pay scale. The host's thesis is sharp: in 2026 cheap technical output is not scarce, judgment under uncertainty and accountability for the outcome is — deciding what not to build and being the name on the failure. Look across the top and pay is concentrating in roles that decide rather than execute. A usage backdrop is added: regular AI use is about 28 percent in the US versus 61 percent in Singapore and 64 percent in the UAE, and running an agent does not even crack the top ten use cases — not because it is hard, but because almost no one has sat down to try it once.

Five shrinking jobs and a practical task lens

The other direction converges on five occupations, all rated very high exposure. Fifth is customer service representatives down about 5 percent for roughly 141,000 lost jobs on a base of 2.6 million; a bank research note finds call-center employment about 39 percent below trend and the oft-cited case is a fintech whose agent handled volume equal to about 850 staff — yet the occupation still posts about 289,000 openings a year, the most of any occupation, so shrinking means a falling ceiling, not disappearance. Bookkeeping, accounting and auditing clerks fall about 5 percent — 85,000 jobs on a 1.5 million base, one of the largest absolute declines, happening one small business at a time — while payroll and timekeeping clerks drop about 15 percent, losing about 25,000 of 159,000 jobs despite paying about 58,000 dollars above the national median, because rules-based, high-volume repeatable work maps cleanly to software. Data entry keyers fall about 25 percent at about 41,000 dollars, losing a quarter of the occupation, and word processors and typists fall about 34 percent — about 36.1 in the official table — the fastest percentage decline in the country for people who type letters and reports from rough drafts.

The close shifts from titles to tasks and borrows a three-column lens from a Stanford economist: every project can be split into defining the question, executing once it is defined, and evaluating whether the result is what you wanted. AI agents are strongest today in the middle column, and the host assigns homework: write down what you actually did over the last two weeks and sort each item into define, execute, evaluate — for a weekly customer report, define what the team needs to learn, execute by gathering data and building the report, evaluate whether the numbers are right and what to do next. Start with one repetitive middle-column task, let AI handle it and check whether it saved time and whether you trust the output; if it works, ask what responsibility you could take on with the freed time — investigating why customers leave instead of just reporting how many left. The host adds a personal entrepreneur test of going away with kids and limited phone access, spotting problems, then building agents on return, and lands where the thesis began: a profession need not disappear for the work underneath to change completely, so choose one task to improve and one responsibility to grow into this week.