Meta's AI Pivot: From FAIR's 2013 Founding Through the Metaverse Detour to $115–145B in 2026 Capex
Summary
- • Zuckerberg published 6,537-word AI manifesto 'The Future is for Everyone' in August 2026
- • Meta founded FAIR in 2013 but lost AI ground during years of metaverse-first strategy
- • Meta's 2026 capex forecast: $115–145B, the majority allocated to AI infrastructure
- • Metaverse division shed 1,000+ roles in early 2026 as structural pivot to AI accelerated
Details
Meta founded FAIR in 2013
Meta founded Facebook AI Research (FAIR) in 2013 as a fundamental AI research lab; made significant open-source contributions including early work on PyTorch
Metaverse pivot cost Meta its AI lead
Post-2021 rebrand, Meta heavily prioritized metaverse and Reality Labs, posting billions in quarterly operating losses while losing ground in frontier AI to OpenAI and Google
Metaverse retreat began early 2026
Meta cut 1,000+ metaverse division roles (approximately 10% of Reality Labs) in early 2026 and began reducing Reality Labs budgets as its AI pivot accelerated
Zuckerberg's AI manifesto, August 2026
Zuckerberg published a 6,537-word manifesto titled 'The Future is for Everyone' in August 2026, outlining a vision for personal superintelligence distributed via open AI models rather than centralized services
2026 capex forecast: $115–145B
Meta forecasts $115–145B in 2026 capital expenditure, with the majority allocated to AI infrastructure and data centers — an unprecedented investment level for the company
Open-weight model releases resumed
The manifesto was accompanied by resumed open-weight model releases including Muse Glimmer, reinforcing Meta's positioning as a champion of open AI
Open AI vs. closed competitor strategy
Meta's open-model strategy directly challenges OpenAI and Google's closed approaches, betting on distributed personal AI agents over centralized AI services
Details compiled via Grok research
Details compiled via Grok live web research from The Verge, TechCrunch, Financial Times, NYT, and Meta's official blog posts
Based on Fast Company reporting by Harry McCracken, supplemented by Grok research digest drawing on The Verge, TechCrunch, Financial Times, NYT, and Meta's blog.
What This Means
Meta's AI trajectory is a case study in strategic overreach and costly correction — a decade of foundational research at FAIR was partially squandered during the metaverse era, and the company now faces the enormous expense of catching up. With $115–145B in projected 2026 capex and Zuckerberg staking Meta's identity on open AI models, the company is making a historic bet that open-weight infrastructure and personal AI agents can compete with OpenAI and Google's closed frontier models. The outcome will shape both the competitive landscape of frontier AI and how AI capability is ultimately distributed across society.