PwC's 2026 Global AI Jobs Barometer does not support the easy panic story. The companies most exposed to AI are adding staff faster, but they're also raising the bar for the people they hire.
The AI-kills-jobs story has been told with great confidence. PwC's latest labour market data makes it look too simple. According to PwC's 2026 Global AI Jobs Barometer, released on June 15 and based on more than one billion job advertisements across 27 countries, headcount at the most AI-exposed companies has grown 52% since 2018, compared with 36% at the least exposed companies.
That is not a small gap. It means the firms closest to AI are not, at least in this dataset, behaving like companies that have found a clean way to replace workers. They are hiring more people and paying them more. PwC found wages at the most AI-exposed companies rose 24% over the same period, compared with 17% at the least exposed companies.
Look, that does not mean every worker is safe. It means the bluntest version of the automation argument is wrong. If you run a company, the useful question is not whether AI cuts jobs. The useful question is what kind of work your AI investment makes more valuable.
The split is already visible
PwC draws a line between what it calls professionalised and democratised roles. The names are dry, but the distinction is useful. Professionalised roles are jobs where AI strips away routine work and leaves humans doing the judgment-heavy part. PwC points to radiologists and recruiters as examples. A radiologist may use AI to filter diagnostic images. A recruiter may use it to automate sourcing, then spend more time on candidates, hiring managers and judgment calls.
Those jobs are doing well. PwC found professionalised roles are growing twice as fast as democratised roles, and their salaries have risen 42% faster since 2021. That is the wage signal you should pay attention to. The market is not rewarding people for merely sitting near AI. It is rewarding people who can use it to do harder work.
Democratised roles sit on the other track. PwC uses IT service managers and medical secretaries as examples of jobs where AI makes more of the work accessible to non-experts. These roles are not vanishing, and it would be lazy to pretend they are. But they are not where the salary pressure is building. When a tool lowers the skill barrier, the worker has less pricing power unless they bring something else to the job.
For founders and operators, this is the part to take seriously. If your AI plan is only a cheaper way to complete existing tasks, you are building into the flatter side of the labour market. If it lets a smaller team make better decisions, handle more complex customers or produce work that previously required a much larger staff, you are closer to the side PwC says is pulling away.
Entry-level work is getting harder, not easier
The uncomfortable finding sits at the bottom of the ladder. Business Insider reported that PwC's analysis of 2.4 million entry-level roles in the US found that AI-exposed junior jobs are now seven times more likely than in 2019 to ask for traditionally senior skills. PwC's list includes leadership, team building, stakeholder management, mentorship and data-driven decision-making.
That is a strange bargain for young workers. AI may remove some of the dull work that used to fill the first few years of a career, but dull work was also how many people learned. PwC found AI-exposed entry-level roles that added more than 10 senior-level skills grew 35% between 2019 and 2025. Comparable roles that did not add those skills fell 10%.
So yes, there is growth. There is also a narrower door. PwC itself is planning to cut US entry-level hiring by about a third over three years, according to Business Insider, and has reduced the number of offices where new consultants can work from 72 to 13. That is not a footnote. It is what this transition looks like inside a firm that is telling clients to adapt to AI.
The sector picture is uneven too. ITPro, citing the PwC report, noted that technology, media and telecommunications companies now account for nearly one in eight new AI-related roles, while professional services follows at 6%. Healthcare sits below 1%, which is less surprising when you remember how slowly clinical workflows move under regulation, liability and patient safety constraints.
The broad labour market number is still striking. AI-specific roles are growing roughly eight times faster than the overall job market, at 69% versus 9%, and PwC puts the wage premium for workers with AI skills at 62%, up from 57% a year earlier. In consumer markets, that premium exceeds 100%.
Also read: Meta secretly sent fake teen accounts to flood rivals' chatbots with thousands of crisis prompts • South Korea is treating nuclear construction speed as an AI infrastructure problem and it's right • Vibe coding has rewritten the rules on who gets to build a startup and billions in VC money are betting it sticks
The better reading is not that AI is harmless. Frankly, that would be as careless as saying it simply deletes jobs. PwC's data points to a sharper divide: companies that use AI to amplify skilled people are growing faster, while workers in roles made easier by AI will have to prove they bring judgment, creativity and human contact the tool cannot supply. If you are building, hiring or entering the market now, that is the real test.
