Mark Zuckerberg laid out Meta’s game plan for “playing to win” against Alphabet and Microsoft in the high-stakes AI arms race. Meta’s secret weapon: its walled garden of data.
“There are hundreds of billions of publicly shared images and tens of billions of public videos, which we estimate is greater than the common crawl data set,” Zuckerberg said on Meta’s earnings call on Thursday. It was a not-so-subtle jab at competitors Google, Microsoft, and OpenAI, which are training their AI models on the public web data crawled by their search engines every day.
Beyond the generative AI technology that Meta already offers in its battle against the other Big Tech powers (including Apple, which teased on its earnings call Thursday that it would launch a generative AI product later this year), Zuckerberg is aiming for “general intelligence,” the still unproven concept of a general purpose AI that can handle most tasks better than humans.
“We’ll be building the most popular and advanced AI products and services. If we succeed, everyone who uses our services will have a world-class AI assistant to help get things done,” Zuckerberg said, even as his management team hinted at the hefty price tag of carrying out the project. Meta’s capital expenditures could increase by as much as $9 billion this year, totalling between $30 billion to $37 billion, compared to $28.1 billion in 2023.
And the company suggested this could be the new normal, noting that “we expect our ambitious long-term AI research and product development efforts will require growing infrastructure investments beyond this year.”
It’s probably no coincidence that Meta announced its first ever cash dividend to investors on Thursday. The new, 50-cent quarterly dividend, along with strong results in Meta’s core advertising business, ignited Meta’s stock, which shot up 14.5% in after hours trading.
Zuckerberg initially unveiled his vision for artificial general intelligence last month in an Instagram Reels video. On Thursday, he expanded on the plan, describing it as not just a far-off science project, but an effort that will incorporate AGI into Meta products like Facebook, Instagram, and Whatsapp.
“It’s clear that we’re going to need our models to be able to reason, plan code, remember, and many other cognitive abilities in order to provide the best versions of the services that we envision,” he said. “We’re playing to win here, and I expect us to continue investing aggressively in this area.”
To realize his vision of Meta as the AI leader, Zuckerberg has yet to expand on how exactly Meta user activity will be used as training data for Meta’s AI, and how the company will navigate the potential issues around user privacy.
And it remains to be seen whether Zuckerberg’s confidence in the value of Meta’s training data compared to the vast corpus of web data used by Google is as big an advantage as he believes. Google has the data treasure trove of YouTube, Google Office, and its search engine. Still, Google AI is closed-source, so it can only move as fast as engineering talent, whereas Zuckerberg pointed to his company’s open-source Llama model as another key advantage.
Earnings aside, this has been a big week for Meta’s head honcho. Yesterday Zuckerberg was skewered during a U.S. Senate hearing where he, most notably, stood up and apologized to families whose children died from harms inflicted by social media. (Though he did not agree to compensate the families for the harms they suffered at Senator Josh Hawley’s request.)
While the apology may have been wrenching for Zuckerberg, his net worth rose by $1.6 billion (per Forbes) on Thursday with Meta’s stock performance.
