When Meta Platforms reports its fourth- quarter results on Wednesday, the questions are likely to go beyond the typical ones about quarterly performance and forward guidance. Investors will want to know CEO Mark Zuckerberg’s thoughts on the new Chinese AI model, DeepSeek R1, and whether his aggressive 2025 capital program remains warranted.
Meta CEO Mark Zuckerberg will likely be asked by Wall Street analysts about his plans for spending in light of a new, apparently more efficient AI model from DeepSeek.
DeepSeek’s achievement was making a good language model and chatbot, like ChatGPT, at a far lower cost than any U.S. contender, or so it claims. If its methods can be replicated by others like Meta, it could dramatically reshape the AI landscape, making AI computing hardware less important and the AI software and services built on top of it more valuable.
“In a world where the cost of intelligence will continue to drop rapidly, more value will accrue back into the app layer,” said Box CEO Aaron Levie in a post on X. “Products that combine AI, customer workflows, and likely some degree of unique data, will generate substantial value from these models going forward.”
So the first question for Meta on Wednesday may be how—with $41 billion of research and development expense in the last 12 months and $30 billion in capital expenditures for AI data centers—an unknown Chinese lab was able to surpass Meta AI in performance at an apparently lower cost.
Even in the wake of R1, Zuckerberg doubled down on Meta’s capex commitments in a Facebook post last week, pledging to spend $60 billion to $65 billion in 2025.
There have already been tough questions asked by investors about Big Tech’s AI spending, but now they’re likely to be amplified.
Amazon, Microsoft, and Alphabet’s Google have spent more on AI data centers than Meta, but those companies have cloud services where they rent out some of their computing power. Meta is buying AI hardware for its own use.
If Meta and others can replicate what DeepSeek has done on the cost side, then the flood of capex may be unnecessary.
There’s a silver lining in the DeepSeek questions, though. If Meta can replicate DeepSeek’s achievements, the data centers it’s already built would become much more efficient.
Meta may represent “one of the largest applications of AI out there, using AI to improve both the user experience and the advertiser experience,” Evercore ISI analyst Mark Mahaney told Barron’s. “So I would think they’d be a major beneficiary of any sort of efficiency hacks.”
Meta shares are up 4.1% so far this week, while the S&P 500 has fallen by 1.5%.
Beyond the effects of R1, Meta also has a fourth quarter to report. According to FactSet, Wall Street analysts expect earnings-per-share of $6.76, up 27% from a year earlier. Sales are projected at $47 billion, up 17%. Those growth rates, while solid compared with many companies, represent a decline from each of the previous five quarters. Analysts are projecting additional declines in growth rates going forward.
Meta is also likely to provide first-quarter guidance for sales, which analysts see at $42 billion, up 14% on the year. Typically, it also discusses operating expenses and capex for the new year. Expenses are expected to rise 14%, while Zuckerberg’s recent capex forecast suggests growth of 63% for the year. Should one of those come in significantly higher, it may pose more questions.
Meta’s stock needs two things from this round of earnings: As always, it will benefit from an earnings beat and a strong forecast. Potentially more important this time around is for Zuckerberg to deliver a convincing defense of Meta’s 2025 expenses.
Comments