A bet sized to the ambition
Meta raised its 2026 capital expenditure guidance to a range of 145 billion, up from the prior 135 billion range, as CEO Mark Zuckerberg laid out what he now calls Meta's organizing mission: delivering personal superintelligence to billions of people.
"We're on track to deliver personal superintelligence to billions of people," Zuckerberg said on Meta's Q1 2026 earnings call, a sentence that would sound deranged coming from almost any other CEO.
The capital plan reflects the spend required to train frontier-scale models, run agentic products at consumer scale, and operate the data center build-out Meta has been constructing across multiple US sites. Total 2026 expenses are projected between 169 billion, a steep jump from $117.69 billion the prior year, driven largely by compensation as the company recruits aggressively for its superintelligence team.
Muse Spark is the first product out the door
The first concrete output from Meta Superintelligence Labs is Muse Spark, a new model that began rolling out in the Meta AI app, on meta.ai, and across Meta's glasses lineup in April and May 2026. Meta says the model took about ten months to ship and has already driven double-digit percent increases in Meta AI sessions per user. Weekly business agent conversations grew tenfold from the start of the year, to more than 10 million.
Zuckerberg drew a sharp line against rivals on the call. Where OpenAI and Anthropic frame AI as a system that automates valuable work, he said, Meta frames it as a system that amplifies individual agency. "I hear a lot of people out there talk about how AI is going to replace people. Instead, I think that AI is going to amplify people's ability to do what they want."
What personal superintelligence actually means
In a public letter last year, Zuckerberg described the bet more concretely. He argued that superintelligence should not be directed centrally toward automating all valuable work, but toward giving individuals more agency to shape their own lives. The intersection of technology and how people live is Meta's focus, he wrote, and will only become more important.
On the latest earnings call, he was more operational. Meta's goal, he said, is not just to deliver Meta AI as an assistant, but to deliver agents that understand a user's goals and work day and night to help achieve them. Enterprise is part of the picture too: APIs, business agents, and possibly selling compute directly to large customers are all on the table, with Meta starting by serving its existing base of advertisers through AI agents that work across messaging apps.
The strategic message is that Meta is no longer competing to ship the best chatbot. It is competing to own the personal AI layer that sits between a user and the rest of the internet, and it is willing to spend the gross domestic product of a mid-sized country to get there.