Meta Delays Avocado AI Model to May, Eyes $115–$135B Capex
Meta delayed its Avocado AI model rollout to at least May from March after performance lagged rival offerings. The company maintains $115–$135 billion capex plan for AI infrastructure and is exploring temporary licensing of Google's Gemini to support its AI product roadmap.
1. Avocado Model Rollout Delay
Meta has postponed the launch of its Avocado AI model from March to at least May, citing underperformance against competitor benchmarks. This delay shifts the timing of expected product enhancements and could affect short-term AI-driven engagement metrics.
2. Underwhelming Performance Metrics
Internal evaluations found Avocado trailing leading models on key benchmarks, prompting additional development and testing. Meta aims to refine model accuracy and capabilities before public release to maintain competitive positioning.
3. Sustained AI Capex Investment
Meta reaffirms its $115–$135 billion capital expenditure plan for AI infrastructure this year, underscoring commitment to long-term superintelligence goals. The spending covers data centers, custom chips and engineering talent expansion.
4. Exploring Gemini Licensing
Leaders within Meta’s AI division are weighing temporary licensing of Google’s Gemini to power immediate AI features. This measure seeks to bridge capability gaps while Meta finalizes its proprietary model improvements.