Meta is to nearly double its capital investments aimed at AI this year, spending more on infrastructure than the entire output of some mid-sized economies, as the AI datacenter feeding frenzy shows no sign of ending.

Facebook's parent company disclosed during a conference call covering its FY 2025 earnings that it expects capital expenditure this year to be in the range of $115 billion to $135 billion, compared with the $72.22 billion it invested in 2025.

This increase is to support Meta's Superintelligence Labs as well as its core day-to-day business, according to chief financial officer Susan Li during an earnings conference call with analysts.

[1]

If it goes as planned, this means that Meta will fork out about as much on infrastructure as the entire gross domestic product (GDP) of Kenya, which stood at $136 billion in 2025, according to [2]figures from the IMF .

[3]

[4]

The social media giant isn't the only tech biz laying out the big bucks to expand its infrastructure. Last year, researchers at Omdia highlighted that [5]Amazon's annual datacenter capex exceeded $100 billion.

Meta believes that this gamble on boosting AI will pay off in its bottom line. "Despite the meaningful step-up in infrastructure investment, in 2026, we expect to deliver operating income that is above 2025 operating income," Li said.

[6]

Founder and CEO Mark Zuckerberg justified all this lavish outlay by claiming that Meta is now seeing a major AI acceleration.

"I expect 2026 to be a year where this wave accelerates even further on several fronts. We're starting to see agents really work. This will unlock the ability to build completely new products and transform how we work," he said on the same conference call.

Meta's pathway to this vision is building what Zuckerberg calls "personal superintelligence."

[7]

"We're starting to see the promise of AI that understands our personal context, including our history, our interests, our content and our relationships. A lot of what makes agents valuable is the unique context that they can see," he said.

Meta is also working on merging large language models (LLMs) with the recommendation systems that power Facebook, Instagram, Threads, and its ad systems, Zuckerberg said, as a way to drive more ad revenue.

[8]Paranoid WhatsApp users rejoice: Encrypted app gets one-click privacy toggle

[9]FTC tries to un-Zuck Meta's grip on the market by dragging it back to court

[10]Meta retreats from metaverse after virtual reality check

[11]Zuck forms Meta Compute to pave the planet with 'hundreds of gigawatts' of AI datacenters

This is where some of the growth in infrastructure spend comes in, as Meta increases the complexity and size of its AI models to better select which ads to show. It doubled the number of GPUs used to train its GEM generative ads model ranking in Q4, for example.

"Today, our systems help people stay in touch with friends, understand the world and find interesting and entertaining content. But soon, we'll be able to understand people's unique personal goals, and tailor feeds to show each person content that helps them improve their lives in the ways that they want," he claimed.

New agentic shopping tools will allow people to find just the right very specific set of products from the businesses in Meta's catalog, he added.

Meta reported revenue of $59.89 billion for Q4 of 2025, up 24 percent year-on-year, while the figure for the whole year was $200.96 billion, up 22 percent.

For the first quarter of 2026, the firm expects its total revenue to be in the range of $53.5 billion to $56.5 billion, slightly down on the Q4 results.

Earlier this month, Meta unveiled a new initiative called " [12]Meta Compute " to oversee its growing network of AI datacenters, saying it plans to build tens of gigawatts of capacity this decade, and hundreds of gigawatts or more over time.

The social media giant also signed agreements with three companies to provide [13]nuclear energy that will be required in future for all the extra datacenters and infrastructure it is planning to build. ®

Get our [14]Tech Resources



[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aXuSffSaJC9w3xhO8DEKrwAAAcE&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0

[2] https://www.worldometers.info/gdp/gdp-by-country/

[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXuSffSaJC9w3xhO8DEKrwAAAcE&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXuSffSaJC9w3xhO8DEKrwAAAcE&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[5] https://www.theregister.com/2025/08/14/datacenter_investment/

[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXuSffSaJC9w3xhO8DEKrwAAAcE&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0

[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXuSffSaJC9w3xhO8DEKrwAAAcE&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0

[8] https://www.theregister.com/2026/01/27/whatsapp_strict_account_settings_meta_rust/

[9] https://www.theregister.com/2026/01/21/ftc_to_drag_meta_back/

[10] https://www.theregister.com/2026/01/16/meta_quest_horizon_workrooms/

[11] https://www.theregister.com/2026/01/12/meta_compute/

[12] https://www.theregister.com/2026/01/12/meta_compute/

[13] https://www.theregister.com/2026/01/09/meta_nuke_deals/

[14] https://whitepapers.theregister.com/