Mark Zuckerberg lays off 8,000, spends $130 billion on AI, then says AI has created a lot of jobs
Meta CEO Mark Zuckerberg has a message for anyone anxious about AI taking their job: the technology is actually hiring. Speaking to the Wall Street Journal hours before Meta’s June quarter earnings, Zuckerberg said that “all the work around AI has net created a lot of jobs because there’s all this infrastructure that needs to get created,” adding that fears of mass displacement haven’t “played out the way that people feared it might.“It’s a claim that invites an obvious question, because the company making it spent this year doing the opposite. Meta laid off 8,000 employees—about 10 per cent of its workforce—in May as part of an AI-first restructuring, reassigned another 7,000 to AI initiatives, and is on track to spend between $130 billion and $145 billion on AI infrastructure in 2026, roughly double last year’s outlay. Zuckerberg isn’t wrong that AI is creating jobs. They just aren’t the ones his own company eliminated.
The jobs Mark Zuckerberg is counting are on construction sites, not inside Meta
His argument rests on the physical buildout. Meta operates or is constructing 32 data centres worldwide and plans to double its computing power to 7 gigawatts this year, then double it again to 14 gigawatts in 2027. Facilities on that scale need construction crews, electricians, cooling engineers and power infrastructure—employment that exists purely because of AI’s appetite for compute. Zoom out and the numbers get larger still: Big Tech’s combined AI spending is projected to cross $700 billion in 2026, with Alphabet guiding to $205 billion and Microsoft to about $175 billion.
Inside Meta, the AI transition looked nothing like a jobs boom
The view from Menlo Park was bleaker. The May 20 layoffs began with 4 am emails in Singapore and rolled westward through Britain and the US; at least one person hired within the previous month was cut. Colleagues marked departures with salad emojis—internal shorthand for “salute”—while offices sat empty after HR told everyone to work from home. The anger predated the cuts. More than 1,000 employees signed a petition against a programme that records keystrokes, mouse movements and screen activity to train Meta’s AI models, with no opt-out on corporate laptops. Those drafted into the new Applied AI and Engineering team were told, in writing, that participation was not optional.
Wall Street is asking Zuckerberg its own version of the jobs question
Investors, meanwhile, want to know what all this spending buys. Meta’s second-quarter free cash flow cratered 91 per cent to $784 million, its lowest since late 2022, and shares fell 10 per cent in extended trading despite revenue climbing 28 per cent to $60.8 billion. Zuckerberg’s pitch to analysts was that the compute itself is becoming a product: Meta is fielding offers to lease capacity “at a significant premium” to what it paid, and is reportedly in early talks with Anthropic. But he resisted the idea of cashing out, telling analysts it “would be foolish” to sell all the compute for short-term profit when Meta needs it to train models and build the personal AI agents he’s betting on. “My personal bet is that the people who invest in this are going to be rewarded,” he said.Both things Zuckerberg says can be true at once. AI is generating work—in Louisiana concrete pours and gigawatt substations. What his sunny arithmetic skips is who bears the cost of the transition: the 8,000 people whose jobs funded the buildout, notified by email before dawn. The concern hasn’t played out the way people feared, perhaps. For some, it played out worse.
