Will Mark Zuckerberg’s AI drive fare better than his metaverse?

The term “metaverse” has faded from public conversations; virtual reality goggles remain a niche product.

Financially, the metaverse lark will go down as one of the most epic wastes of money in corporate history. In nearly six years, Reality Labs accumulated a loss of $77 billion (£58 billion), equivalent to over $1 billion a month.

Which brings us neatly back to today. As we barrel toward 2026, Zuckerberg has dramatically deprioritised the metaverse — he is expected to lay off as much as 30 per cent of Reality Labs staff as soon as next month — to pursue a new obsession: AI, another shiny new technology on which he will spend tens of billions of shareholders’ cash.

• Facebook owner strikes AI deals with news publishers

This raises a basic question: is Zuck steering Meta headlong into another expensive boondoggle? Perhaps more fundamentally, is he any good at making products? At tapping into the zeitgeist? Does he know what people want? The joke, years ago, was that Facebook’s head of research and development was actually Snapchat boss Evan Spiegel, given how fervently Zuckerberg copied his rival’s latest features.

There is no doubt that Zuckerberg is a ruthlessly efficient operator who has leveraged his initial invention into an empire. But as an innovator — a tastemaker — might he be, for want of a better word, clueless? It is a question investors are pondering as Zuck embarks on his next obsession, spending so much of their money to chase it.

Since June, the 41-year-old has spent an estimated $20 billion (£15 billion) on a new unit, Meta Superintelligence. The goal: to build “personal superintelligence for everybody”. What does that actually mean? We’ll come back to that.

The lion’s share of the outlay came through a $14 billion (£10 billion) investment in June in Scale AI, a 1,400-person start-up led by Alexandr Wang. Under the deal, Wang, 28, agreed to join Meta to lead its Superintelligence Labs.

But Zuck was only just warming up his cash bazooka. The numbers are hard to believe. Matt Deitke, a 24-year-old coder, apparently accepted a four-year, $250 million (£187 million) offer from Zuck. Andrew Tulloch, co-founder of Thinking Machines Lab, the AI start-up led by OpenAI émigrée Mira Murati, has reportedly rejoined Meta on a six-year, $1.5 billion (£1.1 billion) pay deal; he worked there as an engineer for 11 years until September 2023.

OpenAI research head Mark Chen recounted how Zuckerberg tried to lure half his workers, offering them untold riches and even hand-delivering homemade soup. “It was shocking to me at the time,” Chen recalled.

Zuckerberg boasted to investors: “We just did this huge effort to boot up Meta Superintelligence Labs and build what I am very proud of is, I think, the highest talent density … in the industry.”

• Meta wooing ChatGPT staff with $100m bonuses, Sam Altman claims

In a recent interview, he said reports on the huge pay packages were inaccurate but admitted it was a “very hot market.” Zuck’s underlying belief is that there are very few difference makers in AI — maybe a few hundred globally. Through that lens, paying vast amounts for the right people is utterly logical if, indeed, the prize is superintelligence. This is a squishy term, but is broadly understood as AI systems which are better than humans at all economically valuable work, and can self-improve and operate with increasing levels of autonomy.

Zuck’s objective is to gift each of us our own personal superintelligence “that helps you achieve your goals, create what you want to see in the world, be a better friend and grow to become the person you aspire to be”.

Wall Street is sceptical — not only of this rather fuzzy vision but the money Zuck is spending. Because he isn’t just blowing cash on people. It’s also on data centres.

Meta has forecast that this year it will lavish up to $72 billion —(£54 billion) nearly three times its budget in 2023 — on capital expenditure.

This is principally going towards high-priced AI chips and gargantuan projects such as Hyperion, a data centre the size of Manhattan that would require up to five gigawatts of electricity — enough to power Greater London.

• Silicon Valley’s AI-fuelled madness has echoes of the dotcom crash

Amid a broader sell-off, Meta’s shares have performed worse than Big Tech rivals, plunging by 13 per cent in the past three months to $644 a share, although that is still up 8 per cent on the year. This is not surprising; building a tool to help us all become our best selves is, to put it mildly, vague.

Zuckerberg put some meat on the bones in April, when he talked on a podcast about how AI companions can combat the “loneliness epidemic” — a crisis that, critics would argue, the owner of Instagram, Facebook, Threads and WhatsApp has helped to create. (Meta vehemently disagrees.)

Most people are faced with a friendship gap. Americans typically have three friends but want far more, Zuckerberg has said. Enter AI, which can magic up any number of anthropomorphised algorithms to be one’s buddy. “I would guess that over time,” he explained, “we will find the vocabulary as a society to be able to articulate why that is valuable.”

Or maybe the answer is more straightforward: AI is just going to make social media more addictive, which means more ads, which means more money. This is the most boring, but also most likely, outcome. Indeed, it seems to already be taking shape. Zuckerberg told investors in October that, in the three months through September, Meta’s AI-powered recommendation systems led to a 5 per cent increase in time spent on Facebook, and a 10 per cent jump on Threads.

“If we deliver even a fraction of the opportunity ahead for our existing apps and the new experiences that are possible, then I think the next few years will be the most exciting period in our history,” Zuckerberg said.

Maybe this AI spending orgy will turn out better than the metaverse. It certainly couldn’t be worse.

danny.fortson@sunday-times.co.uk