Meta CEO Mark Zuckerberg shared a vision in a memo for a near-future with "personal [AI] superintelligence for everyone" forming a new era. The newly formed Meta Superintelligence Labs will develop the next generation of models. This recalls his 2021 Facebook Connect keynote where he laid out the plan for the metaverse.
Zuckerberg believed in the metaverse vision so much that he changed the company name to Meta. But less than four years later, he's now focused on AI "superintelligence" and offering high pay to attract top talent.
Looking back at 2021, the company was heavily selling the metaverse promise. It was supposed to be an "even more immersive and embodied Internet" with various personalized experiences and the ability to do almost anything. But in 2025, the reality of Meta's metaverse efforts is disappointing. It's described as "depressing" and "lonely" with empty venues. The creation of a creator fund also failed to encourage creators.
Fast forward to 2025, Meta's massive investment in the metaverse shows little result with around $60 billion in VR-related losses. But there's still hope with the prototype lightweight augmented reality glasses.
When it comes to AI, Zuckerberg sees it as replacing the current Internet and revolutionizing our lives. It will lead to interactive experiences, improve productivity and creativity, and potentially reduce the need for some employees. However, current AI models still have issues like hallucinating facts and struggling with basic tasks.
There are differences between the past promise of the metaverse and the current promise of AI. Meta's AI products have a large user base, and coders are using AI coding tools. But there are reasons to be skeptical as AI still has limitations. In 2021, Meta executives like John Carmack were upfront about the technical barriers. Now, there seems to be a lack of internal skepticism as Meta pushes AI.