Mark Zuckerberg has a reassuring message for anyone worried about their job: artificial intelligence is hiring. Speaking before Meta's latest earnings, he said all the work around AI "has net created a lot of jobs," and that mass displacement had not "played out the way that people feared." The awkward part is that his own company spent this year doing the opposite. Meta cut 8,000 staff in May, reassigned another 7,000 to AI teams, and is on track to spend up to $145 billion on AI infrastructure in 2026. Zuckerberg is not wrong that AI is creating jobs. They are just not the ones his company eliminated.
The jobs he is counting are real, but they are pouring concrete. Meta's argument rests on the physical buildout: 32 data centres, gigawatts of new computing power, and the construction crews, electricians and cooling engineers needed to raise them. That is genuine employment, and it exists purely because of AI's appetite for compute. What the sunny arithmetic skips is who bears the cost of the transition. A laid-off engineer in Menlo Park does not become a substation electrician in Louisiana. The new work and the vanished work rarely touch the same people, and that gap is the whole story of AI and jobs right now.
Fresh labour data shows the gap widening into two distinct speeds. In the UK, the jobs site Indeed reports that overall postings fell about 11 percent in the first half of 2026, with graduate vacancies at their lowest for the season since 2020, even as mentions of AI skills hit a record 9.4 percent of all listings. Software developer postings actually rose 14 percent, driven by senior and AI-linked roles, while manufacturing vacancies sit 58 percent below their 2022 peak. As Indeed's Jack Kennedy puts it, demand is "concentrating around experienced workers and roles directly connected to AI, rather than flowing evenly." India shows the same split from the sunnier side: white-collar hiring rose 5 percent in July, but AI and machine-learning roles jumped 33 percent, pulling the average up while leaving many sectors flat.
The pattern is unmistakable. Experience and AI fluency are being rewarded; entry-level footholds are disappearing. Britain now has more than a million people aged 16 to 24 not in work or education, and the Bank of England blames a "low hire, low fire" freeze. In China, the rollout of robotaxis has put drivers out of work in Wuhan, and roughly 44 percent of the workforce, some 320 million people, now scrape by in flexible or gig employment. The bar to a first job is rising precisely as the ladder's bottom rungs are pulled away.
Against that, the optimists have their proof points too, and they deserve a hearing. Ikea retrained the call-centre staff whose work its chatbot absorbed, and those centres became its fastest-growing sales channel, expanding 15 to 20 percent a year. The World Economic Forum still projects that 92 million jobs may be displaced by 2030 while 170 million new ones emerge, but only if employers actually invest in reskilling. That "if" is the crux. The Ikea outcome and the Meta outcome both flow from the same technology; what separates them is a deliberate choice about whether to redeploy people or simply release them. AI is not quietly erasing work so much as forcing that choice into the open, and so far the results depend entirely on who is making it.