Tech
Meta's Facebook data center
(George Frey/Getty Images)

Meta is spending more on additional AI infrastructure, and tariffs are pushing the costs even higher

The social media giant raised its 2025 capital expenditures range to $64 billion to $72 billion, from its prior outlook of $60 billion to $65 billion.

Rani Molla

Meta needs more AI infrastructure, so its capital expenditures this year will be higher than it previously guessed. In the company’s latest earnings report, which beat analysts’ expectations, Meta said its 2025 capex bill will now be about $64 billion to $72 billion, a higher and wider range than the $60 billion to $65 billion it had previously forecast.

“We expect this significant infrastructure footprint we are building will not only help us meet the demands of our business in the near term, but also provide us an advantage in the quality and scale of AI services we can deliver,” CFO Susan Li said during the earnings call.

Investors loved Meta’s report, with the stock shooting up 6.8% premarket. Combined with a rosy earnings from hyperscaler Microsoft, the capex announcement also sent AI stocks higher.

The reason for the rising capex bill, though, is not just “additional data center investments to support our AI efforts,” but also “an increase in the expected cost of infrastructure hardware.”

Li wouldn’t break down what portion of the growth was coming from expansion versus higher costs, but did say:

“The higher costs we expect to incur for infrastructure hardware this year really comes from suppliers who source from countries around the world. And there’s just a lot of uncertainty around this, given the ongoing trade discussions. And so that is both reflected in the wider range that we are giving. And we’re also working on our end on mitigations by optimizing our supply chain. And our outlook is really trying to reflect our best understanding of the potential impact this year across all of that uncertainty.”

In other words, tariffs are going to make building AI infrastructure more expensive, since many of the parts and equipment are sourced from places like China. Meta seems game to pay for it anyway.

More Tech

See all Tech
Two Cat Businessmen Holding Drinks

The most outlandish tech CEO quotes from 2025

Tech CEOs have been nuttier than ever.

Rani Molla12/12/25
tech
Rani Molla

Trump AI executive order is a “major win” for Open AI, Google, Microsoft, and Meta, says Ives

President Trump’s new executive order aiming to keep states from enacting AI laws that inhibit US “global AI dominance” is a “major win” for OpenAI, Google, Microsoft, and Meta, according to Wedbush Securities analyst Dan Ives. Big Tech companies have collectively plowed hundreds of billions into the technology, while seeing massive stock price gains, and Ives believes they stand to gain much more.

“Given that there have been over 1,000 AI laws proposed at the state level, this was a necessary move by the Trump Administration to keep the US out in front for the AI Revolution over China,” Ives wrote, adding that state-by-state regulation “would have crushed US AI startup culture.” The presidential order would withhold federal funds from states that put in place onerous AI regulations.

This morning, Whitehouse AI adviser Sriram Krishnan said in a CNBC interview that he’d be working with Congress on a single national framework for AI.

Despite Ives’ rosy read-through on the order, with the exception of Nvidia, which jumped on a report of boosted Chinese demand, many AI stocks are in the red early today. The VanEck Semiconductor ETF is down nearly 1% premarket, as the AI trade struggles thanks to underwhelming earnings results from Oracle earlier this week.

“Given that there have been over 1,000 AI laws proposed at the state level, this was a necessary move by the Trump Administration to keep the US out in front for the AI Revolution over China,” Ives wrote, adding that state-by-state regulation “would have crushed US AI startup culture.” The presidential order would withhold federal funds from states that put in place onerous AI regulations.

This morning, Whitehouse AI adviser Sriram Krishnan said in a CNBC interview that he’d be working with Congress on a single national framework for AI.

Despite Ives’ rosy read-through on the order, with the exception of Nvidia, which jumped on a report of boosted Chinese demand, many AI stocks are in the red early today. The VanEck Semiconductor ETF is down nearly 1% premarket, as the AI trade struggles thanks to underwhelming earnings results from Oracle earlier this week.

tech
Rani Molla

Epic scores two victories as “Fortnite” returns to Google Play and appeals court keeps injunction against Apple

“Fortnite” maker Epic Games notched two wins Thursday in its drawn-out battle against Big Tech’s app stores. “Fortnite” returned to the Google Play app store in the US, Reuters reports, as Epic continues working with Google to secure court approval for their settlement.

Meanwhile, a US appeals court partly reversed sanctions against Apple in Epic’s antitrust case, calling parts of the order overly broad, but upheld the contempt finding and left a sweeping injunction in place — keeping pressure on Apple to allow developers to steer users to outside payment options and reduce its tight control over how apps can communicate and monetize on iOS.

tech
Jon Keegan

Report: AI-powered toys tell kids where to find matches, parrot Chinese government propaganda

You may want to think twice before buying your kids a fancy AI-powered plush toy.

A new report from NBC News found that several AI-powered kids toys could easily be steered to dangerous as well as sexually explicit conversations in a shocking demonstration of the loose safety guardrails in this novel category of consumer electronics.

A report out by the Public Interest Research Group details what researchers found when they tested five AI-powered toys for kids bought from Amazon. Some of the toys offered instructions on where to find matches and how to start fires.

NBC News also bought some of these toys and found they parroted Chinese government propaganda and gave instructions for how to sharpen knives. Some of the toys also discussed inappropriate topics for kids, like sexual kinks.

The category of AI-powered kids toys is under scrutiny as major AI companies like OpenAI have announced partnerships with toy manufacturers like Mattel (which has yet to release an AI-powered toy).

A report out by the Public Interest Research Group details what researchers found when they tested five AI-powered toys for kids bought from Amazon. Some of the toys offered instructions on where to find matches and how to start fires.

NBC News also bought some of these toys and found they parroted Chinese government propaganda and gave instructions for how to sharpen knives. Some of the toys also discussed inappropriate topics for kids, like sexual kinks.

The category of AI-powered kids toys is under scrutiny as major AI companies like OpenAI have announced partnerships with toy manufacturers like Mattel (which has yet to release an AI-powered toy).

tech
Jon Keegan

OpenAI releases GPT-5.2, the “best model yet for real-world, professional use”

After feeling the heat from Google’s recent launch of its powerful Gemini 3 model, OpenAI’s response to its “code red” has been released, reportedly on an accelerated schedule to keep up with the competition.

The company’s new flagship model, GPT-5.2, is out, and the company is calling it “the most capable model series yet for professional knowledge work.”

OpenAI CEO Sam Altman called it the “smartest generally-available model in the world” and shared benchmarks that showed it achieving higher scores than Gemini 3 Pro and Anthopic’s Claude Opus 4.5 in some software engineering tests and abstract reasoning, math, and science problems.

In a press release announcing the new model, the company said: “Overall, GPT‑5.2 brings significant improvements in general intelligence, long-context understanding, agentic tool-calling, and vision — making it better at executing complex, real-world tasks end-to-end than any previous model.”

OpenAI CEO Sam Altman called it the “smartest generally-available model in the world” and shared benchmarks that showed it achieving higher scores than Gemini 3 Pro and Anthopic’s Claude Opus 4.5 in some software engineering tests and abstract reasoning, math, and science problems.

In a press release announcing the new model, the company said: “Overall, GPT‑5.2 brings significant improvements in general intelligence, long-context understanding, agentic tool-calling, and vision — making it better at executing complex, real-world tasks end-to-end than any previous model.”

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, or Robinhood Money, LLC.