Mark Zuckerberg champions ‘superintelligence’ as Meta revenues jump
Meta CEO Mark Zuckerberg is steering the tech giant to invest in creating what he calls “personal superintelligence”, opening its Q2 earnings call and releasing a statement and video pontificating on his new vision for the future of tech.
“We’ve begun to see glimpses of AI systems improving themselves, so developing superintelligence is now in sight”, Zuckerberg declared without providing details.
Meta has established an AI division, “Superintelligence Labs”, aimed at developing the “next generation” of Meta’s large-language models (LLMs).
The division came following Meta’s $14.3bn investment in Scale AI. Taking a 49% stake in the start-up, Meta incorporated Scale AI’s leadership team, including co-founder and CEO Alexandr Wang, within Meta’s leadership structure. Wang now serves as Meta’s chief AI officer.
Meta subsequently went on a hiring spree, recruiting top AI industry talent from competitors like OpenAI, including Shengjia Zhao, the co-creator of ChatGPT, who joined as chief scientist of Superintelligence Labs this month. According to Wired, Meta has offered pay packages to AI researchers of up to $300m over four years.
Zuckerberg has pledged to spend “hundreds of billions of dollars” on developing artificial intelligence products, including building the multi-gigawatt data centres required for their use. Prometheus, the first such centre, is expected to come online next year. It is not clear what the carbon footprint of such efforts is.
The goal, according to Zuckerberg, is to “build personal superintelligence for everyone” that will “help you achieve your goals, create what you want to see in the world, be a better friend, and grow to become the person that you aspire to be.”
Growth despite uncertainty
Such staggering investment in AI is made possible by continued strong growth in Meta’s advertising business, which grew 21% year on year. Total revenue increased 22% to $47.5bn in Q2.
Total ad impressions across Meta’s platforms grew 11% year on year while the average price per ad inflated 9%.
While Meta’s business costs rose 12% to $27.07bn, the tech giant still reported an improved operating margin of 43%.
Meta unveils voice agent and other AI-led features for advertisers
Zuckerberg chalked Meta’s strong quarterly performance up to AI, which he described is “unlocking greater efficiency and gains across our ad system”.
CFO Susan Li also indicated that macroeconomic uncertainty ultimately did not hamper advertiser investment in Q2, including with major “Asia-based online commerce advertisers”. Li noted that “some of the advertisers who cut their US spend in April fully resumed marketing efforts in the US later in the quarter”.
During the quarter, Meta notably introduced ads to WhatsApp, albeit in a limited fashion for now.
Analysis: A ‘metaverse’ redux?
If Zuckerberg’s AI ambitions sound lofty, it is perhaps because they lack detail.
Zuckerberg’s statements call to mind his prior hype for the metaverse, a loosely-defined future where everyone would be using virtual-reality for everything from work to socialising to playing video games.
Meta’s strong growth and high margin gives cover for the tech giant to make long-term investments. Reality Labs, Meta’s metaverse division, lost an additional $4.53bn in Q2 and recorded just $370m in sales. Cumulatively, Reality Labs has lost nearly $70bn since 2020 with little to yet show for it.
Amid increased investor scrutiny, Zuckerberg pivoted away from waxing poetic about unproven business models in the metaverse in 2023, opting for a “back-to-basics” approach that subsequently renewed shareholder confidence.
But given investor interest in AI, and the fact the technology is helping to drive business results for advertisers, Zuckerberg appears to be all-in on the latest hype train.
In his statements, Zuckerberg refrained from defining precisely what he means by creating a “superintelligence” and specifics were scant on how Meta would look to productise any such user application.
“Superintelligence” has generally been used to refer to a hypothetical AI that can cognitively outperform humans in all possible domains of interest. This is distinct from “artificial general intelligence” (AGI), a hypothetical AI that matches human cognitive capacities.
AI experts disagree on whether we’re likely to soon reach AGI, let alone superintelligence, with sceptics warning that tech moguls have embellished the capacity of their AI models.
From a business perspective, Zuckerberg suggested the improved AI would be integrated into its line of smart glasses, but it is not clear how a “superintelligence” version would differ from the company’s extant products, which already incorporate basic AI features.
“I think that personal devices like glasses that can see what we see, hear what we hear, and interact with us throughout the day, are going to become our main computing devices,” Zuckerberg said.
The belief is consistent with Meta’s stance in recent years that “phones have had their time“, as EMEA VP Derya Matras put it as Advertising Week Europe last year.
While Meta’s Ray-Ban smart glasses have proven to be popular (EssilorLuxottica, the owner of Ray-Ban, said revenue from sales of Meta’s smart glasses more than tripled year over year in Q2), they are also far from unseating the smartphone as the core consumer electronic device.
The devices also come with worrying implications for user data privacy, something Meta has a poor history of handling.
We need to talk about Meta — with Outvertising’s Sonnie Spenser