CAREER & HIRING ADVICE

Source check — September 2026. Figures below are current as of the BLS Employment Projections release of August 27, 2026, Forrester’s AI Job Impact Forecast of January 13, 2026, PwC’s 2026 Global AI Jobs Barometer of June 15, 2026, and Challenger, Gray & Christmas data through August 2026.

How many jobs will AI replace by 2030 is not a single census. It is a set of sourced ranges. Treat any page that prints one trophy number, with no method next to it, as a weak brief.

The WEF Future of Jobs estimate is 92 million jobs displaced globally by 2030, 170 million created, net plus 78 million. That is structural change from five macrotrends, not AI alone.

McKinsey is a US hours model, not a pink slip count. The McKinsey America report puts the midpoint at up to 30 percent of hours currently worked in the US that could be automated by 2030. The range is wide, about 3.7 percent to 55.3 percent. Hours automated is not one for one job elimination.

Forrester is the only major house that publishes an actual US replacement headcount for 2030. Its AI Job Impact Forecast, US, 2025 to 2030 puts it at 6.1 percent of US jobs, about 10.4 million roles. The same forecast says AI will strongly influence 20 percent of jobs, 3.25 times the rate at which it replaces them.

BLS does not publish a “jobs AI will replace” headline. The BLS projections still show US employment rising by 5.9 million jobs from 2025 to 2035, from 170.3 million to 176.2 million. Office and administrative support is the major group they expect to shrink the most.

You still have a req open. An engineering manager is waiting on a controls hire. Finance asked if you should pause the seat because “AI is coming for jobs.” The useful work is which seats you keep, which you rewrite, and which you stop posting.

Displacement sits in office support, customer service, food service, and some production, not in a STEM wipeout. Engineering and IT work is more likely to change tasks than vanish. Move mix and screening. Do not freeze a plant seat you already cannot fill.

What do the main forecasts actually say?

They do not say the same thing. WEF estimates global creation and displacement from employer expectations plus ILO employment data. McKinsey models US hours that could be automated, and occupational transitions. Forrester models US structural job loss and augmentation. BCG models how many jobs get reshaped versus eliminated. BLS projects net US employment by occupation and industry.

AI job replacement statistics that mash those into one “replaced by 2030” figure are mixing units. Creation plus displacement is churn. Hours automated is task share. Structural loss is headcount. Net employment is what is left after all of it.

WEF’s method is the Future of Jobs Survey of over 1,000 employers, collectively more than 14 million workers, combined with ILO data. The WEF jobs story names the same survey. The five macrotrends are technology, the green transition, demographics, geoeconomic fragmentation, and economic uncertainty.

Displacement there is a structural change estimate, not a count of people who never work again. On the AI slice, AI and information processing technology is expected to create about 11 million jobs and displace about 9 million. That is one tech trend, not the whole 92 million.

McKinsey’s 30 percent midpoint is the average of a very wide range. Without generative AI the figure was about 21.5 percent. With it, the midpoint moves to about 29.5 percent, an extra eight points, almost 10 percent of US tasks in that framing. The same report estimates about 12 million additional occupational transitions by 2030, with about 11.8 million workers in shrinking demand occupations.

Forrester’s number moved for a reason worth reading. In its 2023 forecast, generative AI accounted for 29 percent of expected US job losses to automation. In the 2026 forecast it accounts for 50 percent, which folds in agentic AI. The headline did not move because models got scarier. It moved because the share of automation attributable to them grew.

BCG’s 2026 model is the clearest statement of the split. Over the next two to three years, 50 to 55 percent of US jobs will be reshaped by AI. Over five years, 10 to 15 percent could be eliminated. Reshaped means the person keeps the job and the daily tools and workflow change. If you only read one sentence before a hiring meeting, read that one.

BLS released Employment Projections for 2025 to 2035 on August 27, 2026. Total US employment is projected up 5.9 million jobs, up 3.5 percent. That program does not publish a single “how many jobs will AI replace by 2030” number. MLR AI case studies show how BLS folds AI into projections. That is still not a replacement census.

Takeaway: the honest answer to how many jobs will AI replace by 2030 is a range with conditions. Globally, WEF’s current figure is 92 million displaced against 170 million created. In the US, Forrester’s 10.4 million is the closest thing to a headcount, McKinsey is talking hours and transitions, BCG is talking reshape versus eliminate, and BLS is still projecting net job growth through 2035.

Skip three viral numbers. Older WEF “85 million displaced” figures are superseded by Future of Jobs 2025’s 92 million. The “400 to 800 million jobs by 2030” line is an older McKinsey global automation scenario, often misread as a 2030 AI replacement census. Goldman Sachs’ “300 million jobs” is an exposure estimate for how many jobs could be affected by automation, not a count of jobs that disappear, and it carries no 2030 date. Any blog that prints it as a replacement total has misread it.

Next move: if a number cannot name WEF, McKinsey, Forrester, BCG, or BLS and the unit it measures, it does not belong in your staffing plan.

Has AI already replaced jobs? What the 2026 numbers show

Every number above is a forecast. There is also a trailing record, and it is the one most pages skip. Challenger, Gray & Christmas has tracked AI as a stated reason for announced US job cuts since 2023. That is employer self-attribution, not verified causation, but it is the only running count anyone keeps.

Through August 2026, AI has been cited in 116,175 announced job cuts, roughly 22 percent of all cuts this year, and it is still the leading stated reason year to date. That is already more than double the 54,836 cuts attributed to AI in all of 2025. Technology leads every industry with 155,126 cuts through August.

The curve is not a straight line up. AI led all monthly reasons for five straight months from March through July. In August it fell to fourth with 3,462 cuts, its lowest month since December 2025. One month is not a trend. It is a reason not to build a five year plan off a single Challenger headline either.

Read the attribution caveat before you quote any of it. Forrester’s J.P. Gownder says that when clients asking for advice on AI-driven layoffs are asked whether they have a mature, vetted AI application ready to fill those jobs, nine times out of ten the answer is no, and they have not started. Most of those layoffs are financially driven and AI is the scapegoat. Challenger’s own framing is the same: the company cited AI, which is not the same as a model doing the work.

Put those two facts side by side. AI-attributed cuts more than doubled year over year, and most of the companies announcing them cannot yet name the system replacing the work. Both are true. The first tells you the labor market is repricing around AI. The second tells you the replacement itself is running behind the announcement.

Next move: when a competitor’s layoff announcement lands in your feed, ask which system took the work. If nobody can name it, that was a budget decision, and the talent it released is worth a call.

How many jobs will AI replace by 2030 in the USA?

How many jobs will AI replace by 2030 in the USA has no official headcount. BLS does not publish that headline. Anyone who prints a single US replacement total as if BLS signed it is ahead of the source.

The closest thing to a direct answer is Forrester’s, and it is worth naming precisely because so many pages quote it without the source. 6.1 percent of US jobs lost to AI and automation by 2030, about 10.4 million roles. Forrester’s own scale check: the US lost 8.7 million jobs during the Great Recession, though the comparison is imperfect because recession losses are cyclical and these are structural.

The augmentation figure next to it is the one that should reach your hiring plan. Forrester expects AI to strongly influence 20 percent of US jobs over five years, 3.25 times the rate of outright replacement, and calls widespread AI-driven job replacement unlikely because labor productivity would have to accelerate far beyond current trends for it to happen.

Those moves are not even. Declines in food services, customer service and sales, office support, and production work could account for almost 10 million, more than 84 percent, of McKinsey’s 12 million shifts.

Lower wage workers in that study, in bands around $38,000 and under, are up to 14 times more likely to need an occupation change than the highest wage group. Women are about 1.5 times more likely to need to move occupations than men. In the MGI Europe update, almost 12 million US transitions is framed as about 7.5 percent of current employment. That is people moving occupations as demand mix shifts, not a stack of empty buildings.

BLS, looking at 2025 to 2035 rather than a 2030 cutoff, still has the US adding jobs. Office and administrative support is projected to shed about 752,100 jobs, down 4.0 percent, the most of any major group. Sales and related is down 1.4 percent. Production is down 0.4 percent.

Computer and mathematical occupations are up 7.3 percent. Professional, scientific, and technical services is up 8.6 percent, about 926,700 jobs, partly from demand for AI systems, R and D, and consulting. Computing infrastructure, data processing, and web hosting is up 25.1 percent, about 120,400 jobs.

McKinsey also estimated about a 23 percent increase in demand for STEM jobs by 2030 in that scenario, even with 2023 tech layoff headlines. A layoff year is not a 2030 demand forecast. Do not freeze a data scientist or controls req because a software company cut staff in 2023.

The sources do not put US plants and IT orgs on the same replacement path as cashiers. The displacement weight sits in office support, customer service, food service, and production mix. Engineering and IT still need people. The work inside the seat changes.

Next move: if finance wants a US number, give them Forrester’s 10.4 million with the 20 percent augmentation figure attached, then hours, transitions, and BLS net growth. Do not pause the controls hire on a global 92 million figure.

What jobs will AI replace by 2030 for engineering and IT teams?

What jobs will AI replace by 2030 is the wrong question if you want a never list. This page is exposure and mix. The jobs AI will replace in the displacement stats cluster where tasks are repetitive, language heavy, and already software wrapped. That is not “your manufacturing engineer is gone.”

WEF asked employers which roles they expect to grow or decline. Fastest declining include Cashiers and Ticket Clerks, Administrative Assistants and Executive Secretaries, Printing Workers, and Accountants and Auditors. Digital access, AI, and robots sit behind a lot of that decline.

Fastest growing tech roles include Big Data Specialists, FinTech Engineers, AI and Machine Learning Specialists, and Software and Applications Developers.

That split should change how you read an engineering org chart. The PLC cell still needs a person. The three administrative seats that grew up around the engineering department are more exposed than the controls lead. An AP clerk pattern in a plant office is closer to the declining WEF list than a process engineer timing a changeover.

McKinsey is blunt on STEM. Generative AI is more likely to change work activities and augment those seats than to wipe out significant headcount. Language models raise task exposure in STEM, creative work, and business and legal. Task exposure is not a headcount zero. If you treat them as the same, you will cancel a software seat you still cannot fill.

BLS points the same direction with different units. Data scientists are projected up 34.6 percent from 2025 to 2035. Computer and information research scientists are up 21.8 percent. Computer and mathematical occupations are the fifth fastest major group.

Production occupations are only down 0.4 percent. That is fewer traditional production seats and more technical work next to the line, not “manufacturing is finished.”

BLS also published AI exposure categories: Low, Moderate, High, Very high. They combined theoretical scores and observed usage. They say an exposure category is not a job loss forecast, not a wage forecast, not an adoption probability, and not a worker replacement estimate. High exposure is not a reason to delete the req.

WEF expects just under 40 percent of workers’ core skills to change by 2030. Human only task share is expected to fall. By 2030, employers in that survey expect a near even split among human, technology, and hybrid work.

Your next software hire is a person who can own a system while a model drafts the boilerplate.

Takeaway: for engineering and IT, the exposure is tasks, adjacent office work, and skill mix. It is not a wipeout of the people who own cells, codebases, and data systems. If you need a list of careers that stay human, that is a different page. Do not turn this section into a second one.

Next move: print your open reqs. Mark each one task exposed or headcount exposed. Rewrite the first. Do not cancel the second without walking the floor or reading the last incident ticket.

What should you watch in staffing plans?

Watch mix, not a hiring freeze. A vacant controls engineer still costs throughput. A model does not climb on the cart next to a cell in fault. If you pause that seat because a blog said AI will replace jobs by 2030, you misread the sources.

Watch what the companies furthest into AI are actually doing with headcount. PwC’s 2026 Global AI Jobs Barometer, built on more than a billion job ads across 27 countries, found the most AI-exposed companies grew headcount faster than the least exposed, 52 percent against 36 percent, and grew wages faster, 24 percent against 17 percent. Jobs requiring AI skills grew 69 percent against 9 percent for the market, with a 62 percent wage premium. The firms best positioned to automate are hiring more people, not fewer.

Watch the junior seat, because that is where the real 2030 change lands. PwC’s US entry level analysis found AI-exposed entry level roles grew 35 percent since 2019 while other entry level roles declined 10 percent, and those roles are seven times more likely to demand traditionally senior skills like judgement and leadership.

AI is eating the routine work that used to function as an apprenticeship. If your junior engineer job description still reads like a list of the tasks a model now drafts, it will attract people who cannot do the part that is left.

That is a rewrite, not a cut. Name what a junior owns end to end in 90 days. Name who reviews their work and how often. If the only thing that seat did was produce first drafts, you do not have a junior role, you have a bottleneck you were staffing around.

Watch the seats around the engineer. If the plant has always had three administrative coordinators in engineering, that pattern is closer to WEF’s declining clerical list and to BLS office support than the manufacturing engineer is. You may need fewer of those seats. You still need the person who owns the line metric.

Watch how you write the req. “Software engineer, competitive pay, ASAP” will attract fluent noise. Name the system, the language, the cloud, the on call, and what they own in 90 days.

Tool use is a skill in the brief. It is not a reason to cancel the search. How you recruit employed engineers does not change because a model can draft a cover letter. That sequence lives on our how to recruit page.

Watch the screen. You are in a slate review. Three resumes look clean. None of the three can name the last system they owned. The engineering manager is waiting.

AI did not replace that seat. A weak screen did. Put interview questions on the work, not on a vibe. Put reference questions on what the person actually shipped.

Watch who you call a partner. Apollo Technical is a specialized engineering and IT staffing firm. The live product list is on the engineering staffing page.

The model is pay on hire. You pay only if you hire. We do not run retained searches. We do not invent a fee percent in an article. If the seat is a working engineer or an IT hire with an active market, that is the product. If the seat is a confidential head of engineering, that is a different firm.

Next move: write one page before anyone sources. Title a peer would use. What they own in 90 days. Must haves, 3 to 5. Salary range, not “competitive.” Interview days. How many firms are already on the req. Then fill the seats the sources still say you need.

Will AI replace jobs by 2030 or change them?

Will AI replace jobs by 2030? Some jobs, in some groups, under some adoption paths. Most of the sourced picture is change, churn, and mix, not a single wipeout year.

Every house that measures both sides lands in the same place. Forrester: 6.1 percent replaced, 20 percent strongly influenced. BCG: 10 to 15 percent eliminated over five years, 50 to 55 percent reshaped in two to three. WEF’s net is plus 78 million globally, with the AI and information processing slice at about 11 million created and 9 million displaced. Change outweighs replacement in every one of them, by a wide margin.

McKinsey is hours and occupational moves, not a pink slip census. BLS still adds US jobs from 2025 to 2035.

The condition next to every yes is adoption. A plant that buys a license and does not change standard work will not see McKinsey’s midpoint. A company that turns every ticket into a prompt and never checks ownership will hire fluent people who cannot debug. Forrester’s own read is that replacing human talent at scale would require labor productivity to accelerate well past current trends. That has not happened yet.

Next move: pick three open seats you will rewrite this quarter because the task mix changed. Pick one seat you will stop posting because the work is already gone. Do not freeze the whole plan.

How is this different from AI proof career lists?

This page is quantity, method, and staffing mix. It is not a list of careers that stay human.

If you want that list, we already published AI proof careers. Use it for career shape, judgment, physical presence, and work that does not sit in a repeating prompt. Do not copy those keywords onto this URL. Do not turn “what jobs will AI replace by 2030” into a second never list.

A controls engineer is not a cashier. An administrative assistant in the engineering pod is not a data scientist. Use this page when finance asks how many. Use the other when a candidate asks what stays human. Your req still needs a title, a salary range, and a person who can own the work.

Short answers hiring managers keep asking

How many jobs will AI replace by 2030?

There is no single census. WEF’s current global estimate is 92 million displaced by 2030 against 170 million created, net plus 78 million, from five macrotrends, not AI alone. In the US, Forrester forecasts 6.1 percent of jobs lost, about 10.4 million roles. McKinsey is US hours and transitions. BLS has no replacement headline.

How many jobs will AI replace by 2030 in the USA?

Forrester’s AI Job Impact Forecast puts it at 6.1 percent of US jobs, about 10.4 million roles, and expects AI to strongly influence another 20 percent. BLS does not publish that number at all. McKinsey’s usable US figures are up to 30 percent of hours automated at the midpoint, and about 12 million occupational transitions. US net employment is still projected up 5.9 million jobs from 2025 to 2035.

Has AI already caused job losses in the US?

Yes, by employer attribution. Challenger, Gray & Christmas counted 116,175 announced US job cuts citing AI through August 2026, about 22 percent of all cuts this year, against 54,836 for all of 2025. The caveat matters: that is the reason companies gave, not verified causation, and Forrester reports that nine out of ten clients planning AI-driven cuts have no vetted AI system ready to do the work.

What jobs will AI replace by 2030?

Displacement stats concentrate in office support, customer service, food service, some production, and clerical roles WEF employers expect to decline. For engineering and IT, use task exposure and adjacent office mix, not STEM headcount wipeout.

Will AI replace jobs by 2030?

Some, in exposed groups, if adoption actually lands. BCG puts elimination at 10 to 15 percent of US jobs over five years against 50 to 55 percent reshaped in two to three. The sourced center of gravity is change and occupational moves, not a year when work ends.

Should you freeze engineering hiring because of AI?

No. BLS still has computer and mathematical occupations growing. McKinsey still has STEM demand up about 23 percent by 2030 in that scenario. PwC found the most AI-exposed companies grew headcount 52 percent against 36 percent for the least exposed. Freeze the weak admin pattern if the work is gone. Do not freeze the controls hire.

Does BLS forecast jobs AI will replace?

No. BLS projects net employment and publishes relative AI exposure categories. Exposure is not a worker replacement estimate.

If you want a pay on hire search for the engineering and IT seats that still need a person, bring the title, the salary range, the must haves, the interview dates, and whether a second firm is already on the req. If the seat is a confidential leadership hire, go to a retained firm. Do not pick a partner by a viral displacement number.