Meta has pushed back the release of its next-generation AI model, Avocado, to at least May, citing performance shortcomings compared to rivals like Google's Gemini 3.0. The delay, confirmed by sources, follows internal tests showing Avocado underperformed against established leaders in reasoning, coding, and writing tasks. This setback underscores the intense pressure Meta faces in the AI arms race, where giants like Google and OpenAI dominate cutting-edge capabilities. Zuckerberg's billions in investment and data center spending aim to close the gap, but the Avocado delay signals a significant hurdle.

The model's performance gap risks undermining Meta's ambition to be a leader, potentially impacting its advertising revenue and ability to attract top AI talent. Executives are now debating licensing Gemini temporarily, though no decision is finalized. Meta's strategic gamble hinges on rapidly advancing its AI, a core pillar of its future growth strategy.