Meta Platforms is considering joining forces with its largest competitors in the race to develop artificial intelligence. According to reports, the company has held talks with Google and OpenAI to integrate their AI models into Meta’s growing suite of AI products. Executives inside Meta Superintelligence Labs, the company’s new AI division, have discussed incorporating Google’s Gemini model. The aim would be to improve what are already highly accurate and natural text-based responses in Meta AI, the company’s leading chatbot, when users ask questions. The discussions also included the potential use of OpenAI’s models, the technology that could enhance AI features across Facebook, Instagram, and WhatsApp. Any partnership would not mean a permanent shift away from Meta’s work. According to people knowledgeable about discussions, these deals are expected to be a stopgap for Meta’s in-house technology to develop. The company’s priority is developing Llama 5’s new large language model . Meta intends for this model to challenge OpenAI’s GPT series and Google’s Gemini head-on. This pragmatic stance conveys the sense of emergency in the AI arms race. Rivals have already introduced sophisticated products, and Meta does not want to lag as it develops its systems. In the short term, borrowing strength from Google or OpenAI makes its AI systems competitive and engaging for users. Meta couples external AI with in-house work As it aims to push toward self-reliance, Meta is already experimenting with AI outside the company. A coding assistant using Anthropic’s models is available to employees who can quickly compose and test software. That shows that Meta is willing to blend outside technology with its in-house tools when doing so, offering a differentiated benefit. A Meta spokesperson emphasized this adaptable strategy, saying the company was taking an “all-of-the-above approach.” That involves creating best-in-class models itself, working with other leading players, and open-sourcing technology. Meta has been among the more aggressive companies in outsourcing its models, releasing earlier versions of Llama to researchers and developers. The push is being fueled by massive investment. Earlier this year, Meta pledged billions of dollars to grow Meta Superintelligence Labs. Mark Zuckerberg enlisted Alexandr Wang, once the chief executive of Scale AI, and Nat Friedman, a former chief executive at GitHub, to co-lead the project. The lab has also recruited a dozen researchers, many of whom were offered very high salaries, keeping with Zuckerberg’s ambition to turn Meta into a leader in AI. The desired angle is clear. Meta wants to be out in front of the AI revolution, not a fast follower. A temporary alliance with OpenAI or Google could bridge some of these holes in the near term, but the company is betting its homegrown technology will eventually rival the competition. Competition shapes Meta’s choices Meta’s approach is similar to, but also different from, how other tech giants deal with the AI arms race. Microsoft was an early bettor on OpenAI, pouring billions into the company and integrating GPT models deeply into its products. Apple has chosen a more balanced route, collaborating this year with OpenAI to integrate ChatGPT with the iPhone and develop its own Apple Intelligence system. Meta’s approach falls somewhere in between. Like Apple, it wants flexibility and will mix partnerships with its models if it gets that. Like Microsoft, it is willing to spend big to attract talent and speed up research. What makes Meta different is that it’s willing to let the world borrow parts of what it built, a strategy that wins friends and influences developers, and applies pressure on closed rivals. For users, that could mean more powerful AI characteristics for its social-media apps, which serve billions. For the industry, Meta’s negotiations with Google and OpenAI shows the fact that even bitter rivals may have to work together sometimes. Cooperation, however fleeting, could be crucial for staying competitive in the AI race. But in the long term, Meta’s fate lies with Llama 5 and beyond. If its models can rival or exceed Gemini and GPT, Meta may not need outside help anymore. Until then, deals with rivals could be the bridge that keeps it in the fight. If you're reading this, you’re already ahead. Stay there with our newsletter .
