Mark Zuckerberg has not lost his sleep in the prospect of the AI bubble. In fact, he has been shrugging more or less on the idea of an Access Podcast this week, saying that while the tech industry may be heading towards one, there are big mistakes sitting there. “AI bubbles are very possible,” admitted Zuckerberg, pointing to the previous boom and bust cycles where businesses jumped out, went down and left the infrastructure that others scooped up afterwards. However, in his view, such waste is still preferred to be left in the dust.
“I think it's obviously very unfortunate if you misinterpret tens of billions of dollars,” he said. “But what I say is, I actually think the risk is high on the other side.”
“The other side” means a nightmare scenario. The world, where artificial tension appears earlier than expected, and despite its size, the meta is captured flat. “The risks for companies like Meta at least are probably not too aggressive, not too aggressive,” he explained.
It's very aggressive. Meta has pledged to cultivate at least $600 billion into US data centers and infrastructure by 2028. Meta's CFO Susan Lee later revealed that not only the blink server rack, but all of Meta's US business operations pledges have been folded. Still, it emphasizes its determination to outperform Zuckerberg's rivals in terms of investment size.
All this comes as a chatter about the AI bubble reaching the heat pitch. Experts are already pulling the line back to the 2000 dot-com crash, when enthusiastic spending clashes with the hard reality. Some investors are worried that history is about to repeat itself. But for Zuckerberg, being ti-disease is a much bigger gamble.
Meta tightened his belt elsewhere. After seducing AI experts with a luxurious signature bonus, the company suddenly suspended last month in recruitment. Wall Street was beginning to question whether the high mountains of labor and stock compensation in the sky were justified, especially if the promised breakthrough remained over the horizon.
Still, the contrast between Meta and its closest AI rival is impressive. Zuckerberg emphasized drawing that line in the podcast. Unlike Openai and humanity, where hats must be handed over continuously to cover huge calculation bills, Meta has a deep pocket luxury.
“We don't risk going out of business,” he said flatly. He noted that private labs rely on both investors' appetites and the broader economy. A recession caused by a global shock could scramble them. “If you're in one of their shoes, that could be a different situation,” he added.
Zuckerberg also lifted the curtains of an approach to building towards super intelligence in the meta. Rather than loading the body into a vast team, the company has established a powerful, lean lab with staffed elite researchers. This group works in the absence of flat structures and strict deadlines, reflecting the unpredictable nature of frontier AI research.
He said the other major differentiator is calculation. Meta wants to have researchers access to more raw processing power than anyone else. In other words, GPUs and custom chips are perks for the new company, aimed at giving Meta teams an edge over their rivals.
For now, Wall Street seems to be buying stories. Meta's stock has grown nearly 40% over the past year, suggesting that investors are happy to plague costs in anticipation of future payments.
Zuckerberg's message is clear whether it's bubble or not. The company wants to swing a bigger and more overshoot than risking lacking what is called “the most important technology that enables the creation and history of the most new products and innovation and value.”
And what if hundreds of billions disappeared in the process? Well, that's just the price of admission.
– end
