Zuckerberg says Meta weighed a $200 billion AI bet before models were proven

Mark Zuckerberg said Meta considered committing roughly $100 billion to $200 billion to AI before executives knew whether its new models would succeed. In an October 12 interview, he recalled discussions from late 2025 with chief AI officer Alexandr Wang. Zuckerberg did not define a spending period or precise scope for those figures.

Wang said early research results encouraged him, but success remained uncertain. He recalled finance chief Susan Li repeatedly warning during an executive gathering in Hawaii that a failure would have severe financial consequences.

The executives described the difficulty of ordering computing infrastructure years before the products it will support are ready. Zuckerberg said Meta had to estimate future demand before those products existed, and chose to continue investing despite arguments for greater caution. His account describes the reasoning behind the investment rather than a newly announced $200 billion purchase.

The discussion coincided with Meta's introduction of Muse Wonder and Muse Wonder Full on October 12. Meta made Wonder available through its model API and Muse Code, with Wonder Full available to Muse Code subscribers. The company says Wonder will next expand into its Muse service, beginning with subscribers, and that further safety evaluation results will follow.