Read: A shame: Meta forces us to offer it our content to train its artificial intelligence
In the second quarter, Meta benefited from advertising sales based on Reels, the short, entertaining videos copied from TikTok. “There was an alignment of the planets between the increase in opportunities for brands and the increase in average ad prices,” commented Max Willens of Emarketer. “With such healthy margins, Meta investors should be comfortable with the company’s significant investment in its future plans.” Even though the Reality Labs division, responsible for developing devices and software for the metaverse (mixing real and virtual worlds via high-tech glasses and headsets), again widened its losses, to $4.5 billion (3.9 billion Swiss francs) this quarter.
Its shares were up more than 7% in electronic trading after the close of trading on the New York Stock Exchange on Wednesday.
Ambitious spending on AI
The market is currently particularly interested in the spending of large technology groups on generative AI – often considered too high – and the potential returns on investment – never fast enough.
But “Meta has earned investors’ patience by investing in its future vision,” said Jasmine Enberg of Emarketer. “They are more comfortable with AI spending (…) because the ad business is strong.”
The Californian group, which was penalized on the stock market in the first quarter due to increased spending, has again raised the range of its capital investments, now between 37 and 40 billion dollars for the year.
Read also: Haven’t been able to keep up with the mad race to AI? Here’s what you need to know
Since the success of ChatGPT, tech giants have been rapidly deploying models and applications capable of producing high-quality content based on a simple query in everyday language. But these models require new IT infrastructures, a lot of energy, highly qualified engineers, etc.
Meta capitalizes on AI
Meta unveiled in April the new version of Meta AI, its assistant that answers users’ questions, like ChatGPT. It has gained visibility on the group’s platforms and skills thanks to Llama 3, the latest version of Meta’s AI model, comparable to GPT-4 (OpenAI) and Gemini (Google).
“Meta AI is well on its way to being the most widely used AI assistant by the end of the year,” Zuckerberg assured analysts during the conference. The amount of computing resources needed to train Llama 4 “will certainly be ten times greater than those used for Llama 3,” he acknowledged. And in the longer term, it is difficult to predict the exact needs, “but given the time to go live, I would rather take the risk of building capacity up front rather than too late,” he specified.
Microsoft (OpenAI’s first investor) and Google are in the lead, but the world’s number two in digital advertising wants to become “the world’s leading AI company.” With significant assets: more than 3.2 billion people use at least one of Meta’s applications on a daily basis.
“And unlike Google, which is facing changes that will impact its core business, most of Meta’s investments in AI are improving how advertising works on its platforms, or creating new features that could become revenue streams,” said Debra Williamson of Sonata Insights.
Meta is already capitalizing on AI with its content recommendation and ad targeting algorithms. “Our systems predict better than advertisers who their ads will interest,” Mark Zuckerberg said. Generative AI will soon create ads itself and personalize them based on the user, he continued. And “in the longer term, brands will give us their business objective and we will do the rest for them.”
See also: In video: The race for general artificial intelligence
