In the AI agent competition, companies like OpenAI, Anthropic, Microsoft, NVIDIA, Google, and Amazon are taking the lead. Meta, despite its successful LLM family, struggles to stay relevant.
Meta’s AI strategy focuses on openness, scale, and control, with its new model ‘Avocado’ facing delays due to performance concerns. This has sparked a debate about open source and profitability.
Meta has accelerated its AI efforts with Meta AI. Initially launched as a chatbot on WhatsApp, Instagram, Facebook, and Messenger in September 2023, it evolved into a standalone app by April 2025, offering a Discover Feed, voice capabilities, and personalization features.
Meta AI, powered by LlaMa, Meta’s LLM family, aims to democratize AI model access. LlaMa launched four models as open-source, and Meta also introduced a limited preview of LlaMa API for developers.
Reports indicate Meta is working on next-gen AI models like ‘Avocado’. Unlike LlaMa, ‘Avocado’ will be proprietary, restricting access to its software components.
Meta’s shift from open-source to proprietary models challenges its previous stance on openness in AI development. This shift is attributed to the rise of competitors like DeepSeek and the strategic need to economize on AI investments.
In June 2025, Meta invested $14.3 billion in Scale AI for a 49% stake, appointing Scale AI’s founder to lead Meta Superintelligence Labs to develop ‘Avocado’.
Meta’s AI strategy faces challenges, especially with LlaMa 4’s ‘Behemoth’ model and the delayed ‘Avocado’ launch. These delays arise from performance issues compared to competitors like Google’s Gemini models.
There are also discussions about Meta potentially licensing Gemini from Google for ‘Avocado’, highlighting a possible dependency on external technology.
Meta’s recent challenges raise questions about its long-term AI strategy. While LlaMa positioned Meta in the open-source scene, leveraging AI investments and addressing execution difficulties becomes crucial.
Meta’s decision to shift from open-source to proprietary models is a strategic move to manage the costs involved in AI development. Delays and underperformance magnify the need to catch up in AI advancement.
Meta facing potential technology dependency on Google indicates a shift from developing core capabilities to distributing AI products. This raises doubts about Meta’s ability to maintain a strong AI strategy, impacting its leading position in the AI industry.