Key Takeaways:
Meta Platforms Inc. is set to release its first major artificial intelligence model under new AI chief Alexandr Wang, a critical test for the company as it invests billions to catch up with rivals like OpenAI and Google. The new family of models, code-named "Avocado," represents a strategic shift for Meta, which has previously focused on open-source releases but is now developing more powerful, proprietary technology to close a significant performance gap with competitors.
The move follows a string of underwhelming AI product launches from Meta, which has struggled to gain traction despite a planned $600 billion investment in the field. "Meta knows its new models may not be competitive across the board with the coming ones from those labs, but believes it will have areas of strength that appeal to consumers,” according to sources cited by Axios. The company's previous Llama series, while technically open-source, failed to capture significant market share from closed-source leaders like OpenAI's GPT series.
The "Avocado" model was reportedly delayed from a planned March 2026 release after it underperformed on key benchmarks for coding, reasoning, and writing. This setback highlighted the immense challenge Meta faces in competing with Alphabet Inc.'s Google and Microsoft Corp.-backed OpenAI, which have established a strong lead in the generative AI market. The new models are expected to prioritize consumer-facing features, including advanced shopping tools, as Meta seeks a viable path to monetization.
For investors, the launch is a pivotal moment. Meta's stock, which has been under pressure due to the high cost of its AI and metaverse bets, hinges on the success of these new models. The company's ability to deliver a competitive product could validate its massive spending and reshape the competitive landscape. Failure, however, would raise serious questions about its AI strategy and the leadership of Wang, who was brought in to turn the tide after Meta acquired a $14 billion stake in his firm, Scale AI.
While Meta has been the most prominent U.S. company to offer open-source frontier models, there is growing speculation that this strategy could be reversed. The company plans to initially release the new models as closed-source to manage safety risks, a departure from its previous approach with the Llama series. This hybrid strategy could involve offering licenses to firms that want to use the model, a different business model than the subscription services favored by its competitors.
The debate between open-source and closed-source AI development is a central theme in the industry. Open-source models, like Moonshot AI's Kimi 2.5, have been used by other companies to build their own products, demonstrating a viable path for widespread adoption. However, leading firms like OpenAI, Anthropic, and Google have recently started collaborating to prevent the copying of their models, particularly by competitors in China, highlighting the risks of a fully open approach.
The success or failure of Meta's new AI push rests heavily on the shoulders of Alexandr Wang. The founder of Scale AI, a data-labeling giant, Wang was brought in to lead a new team focused on artificial general intelligence. His vision centers on creating a "personal superintelligence" that understands a user's goals and interests, a concept that could unlock new advertising and e-commerce opportunities for Meta.
However, the pressure is immense. The delayed launch of "Avocado" and rumors of friction between Wang and CEO Mark Zuckerberg have put Wang in the hot seat. The upcoming release is seen as a sink-or-swim moment for the young executive. If the new models fail to impress, Wang will likely be the scapegoat for Meta's continued struggles in the AI race, while a success could finally establish Meta as a credible contender.
This article is for informational purposes only and does not constitute investment advice.
