Zuck's Super Intelligence Master Plan Revealed

Meta's Aggressive Super Intelligence Play 00:00

• Mark Zuckerberg initiated an aggressive recruitment strategy, targeting top AI researchers from companies like OpenAI and Anthropic.
• Zuck compiled a personal list of top AI talent and offered extremely high compensation, reportedly $100 million signing bonuses.
• The formation of Meta's "super intelligence team" followed the release of Llama 4, which was good but not considered enough to win the AI race.
• Meta acquired a 49% stake in Scale AI for $14 billion, primarily for its data, team, and leadership under Alexander Wang.
• Regulatory and antitrust concerns, along with a desire for speed, influenced Meta's decision not to acquire Scale AI outright.
• After Meta’s deal, OpenAI and Google ended their contracts with Scale AI; Meta was unbothered as they had already secured critical assets.

Major Talent Acquisition and Industry Response 02:43

• Meta shifted Alexander Wang from Scale AI to lead the super intelligence team.
• Meta poached a significant number of top researchers from OpenAI, including the entire Zurich office responsible for key model developments.
• Notable recruits include Shenzha Xiao, Xia Hui Yu, Hong Yu Ren, and Shu Chow B, all leading or contributing to major AI advancements at OpenAI or DeepMind.
• In total, at least eight OpenAI researchers have reportedly joined Meta.
• OpenAI responded with internal memos, acknowledging the brain drain and promising to recalibrate compensation offers to retain talent.
• Meta leveraged OpenAI’s company-wide week off to increase pressure on staff for quick decisions on Meta offers.

Strategic Shifts and Company Priorities 06:11

• OpenAI leadership communicated the need to retain fairness in compensation while striving to keep top talent.
• Meta’s larger resources and public status enable it to make more aggressive offers than OpenAI, which remains a (comparatively) smaller private company.
• OpenAI was criticized for focusing too much on incremental releases, losing sight of the ultimate goal—super intelligence.
• Mark Zuckerberg and Meta emphasize achieving super intelligence above all else, viewing competition with OpenAI as a secondary concern.

Official Launch of Meta Super Intelligence Labs 07:27

• Mark Zuckerberg internally introduced the Meta Super Intelligence Labs (MSL), encompassing all foundation models, products, and research teams.
• Alexander Wang is now Meta’s Chief AI Officer and head of MSL, with GitHub’s Nat Friedman co-leading in applied research and product focus.
• An impressive roster of AI researchers from OpenAI, Anthropic, and Google has been assembled for MSL, including:
  • Trapit Bonsel (reinforcement learning, chain-of-thought models, OpenAI)
  • Shu Chiao Bai (GPT-4o, multimodal post-training, OpenAI)
  • Huen Chang (GPT-4o image generation)
  • G Lynn (multiple GPT mini models, image generation)
  • Joel Pobar (Anthropic, inference)
  • Jack Ray (Gemini reasoning, Google DeepMind)
  • Hong Yu Ren, Johan Schulwick, Pay Sun, Jihu Yu, and Shanggha Xiao (contributions to major model launches at OpenAI and Google)
• The team is described as an "all-star" group working with the sole purpose of developing super intelligence.