Meta Platforms Inc. CEO Mark Zuckerberg is intensifying his personal involvement in the company’s artificial intelligence efforts as Meta pivots toward a new, revenue-focused model codenamed Avocado, according to sources cited by Bloomberg. The AI system is expected to debut in spring 2026 and may be released as a “closed” model—tightly controlled and designed for commercial access—marking a departure from Meta’s long-standing open-source strategy.
The move follows the underwhelming response to Llama 4, Meta’s previous open-source AI model, which led Zuckerberg to restructure the team, recruit top AI talent with multiyear pay packages, and create a specialized group called TBD Lab to focus on the new project. Chief AI Officer Alexandr Wang, brought in through a $14.3 billion investment deal with his startup Scale AI, is leading the development, with Zuckerberg mentoring him closely.
The Avocado team is reportedly training the model using a combination of internal resources and third-party technologies, including Google’s Gemma, OpenAI’s gpt-oss, and Alibaba’s Qwen model. This signals a notable shift for Zuckerberg, who previously expressed concerns over potential censorship in Chinese AI models.
Meta’s AI push has become the company’s top priority, commanding significant investment. Over the next three years, Zuckerberg has pledged $600 billion in U.S. infrastructure spending, largely focused on AI, and is redirecting resources from virtual reality and metaverse projects toward AI-related hardware such as AI glasses.
While investors remain cautious over Meta’s high-cost AI strategy, Zuckerberg insists that these investments strengthen the company’s core advertising business and are essential for “front-loading” AI capacity. However, internal reports suggest the AI division has experienced staff departures, layoffs, and some tension over micromanagement, as Zuckerberg closely monitors progress.
The shift to closed models and de-emphasis of open-source initiatives has also led to high-profile departures, including AI pioneer Yann LeCun, who left Meta citing resource constraints and strategic disagreements, according to Bloomberg.
Meta’s new AI ambition aims to achieve “superintelligence”, systems capable of outperforming humans in a wide range of tasks. While the vision is ambitious, internal and external experts have expressed concerns over regulatory scrutiny and public perception, particularly in Europe, where fears of unchecked AI development are rising.
Bloomberg reports that success for Avocado will hinge on Meta’s ability to balance technological innovation, regulatory compliance, and public trust, as the company navigates one of the most expensive and closely watched AI endeavors in the tech industry.
