Meta, formerly known as Facebook, has announced its decision to charge significant cloud computing companies, such as Microsoft and Alphabet, to use its artificial intelligence (AI) model, Llama 2. This is a surprise move, considering Meta had initially positioned Llama 2 as an open-source tool accessible to everyone. The CEO of Meta, Mark Zuckerberg, revealed that the decision to levy a fee on these companies is based on the idea that they are essentially reselling Meta’s services. Thus, a portion of the revenue should go to the company. This announcement was made during the company’s earnings call on July 26, 2023.
The shift from open-Source to paid model
Llama 2, known as Meta’s large language model (LLM), has gained significant attention for its language processing capabilities, drawing interest from tech giants like Microsoft and Amazon. Originally launched as an open-source product, Llama 2 was meant to be freely available for commercial use on various cloud offerings. Meta’s partnership with Microsoft and Amazon aimed to make the AI model accessible to developers on their platforms.
During the recent earnings call, Zuckerberg explained that the decision to charge fees to major cloud computing companies stemmed from their role as resellers of Meta’s services. According to Zuckerberg, companies like Microsoft, Amazon, and Google, which offer cloud computing solutions to a vast user base, profit from reselling Llama 2 as part of their platforms. As a result, Meta believes that a share of the revenue generated from these resellers should rightfully return to the company.
While Zuckerberg acknowledged that the immediate revenue from these fees might not be substantial, Meta is optimistic about the long-term potential of this business arrangement.
License requirement for companies with large user base
Meta has also introduced a licensing requirement for companies with over 700 million monthly active users. Such companies must obtain a license from Meta to access and use Llama 2 for their AI-related needs. This move suggests that Meta is looking to regulate and potentially monetize the usage of its AI model by large-scale companies with extensive user bases.
Llama 2’s reach expands to China
Before the announcement of charging fees, Meta had expanded Llama 2’s accessibility by partnering with cloud computing providers like Microsoft and Amazon. On July 18, 2023, Meta revealed that Llama 2 would be available in the Azure AI model catalog, allowing users to access the AI model on Windows and through Amazon Web Services (AWS) and other providers.
Additionally, Alibaba’s cloud computing division recently introduced a solution for the entire Llama 2 series in China. This move enabled Chinese developers to create customized large AI models on Alibaba Cloud, further extending the reach of Llama 2 in the global market.
Maintaining openness amidst monetization
Zuckerberg clarified that despite the decision to charge fees, Meta intends to keep Llama 2 as an open-source product. Developers and users will still be able to access the AI model. Still, significant cloud computing companies must enter into a business arrangement with Meta to use the model in their services. The CEO emphasized that maintaining the open-source nature of Llama 2 is crucial for fostering innovation and enabling developers worldwide to benefit from the AI model’s capabilities.
Meta’s surprise announcement to charge major cloud computing companies for using Llama 2 marks a significant shift from the model’s initial open-source approach. While it aims to regulate revenue sharing from reselling services, Meta is also committed to openness, ensuring that developers can continue to access the AI model for various applications. As the AI landscape evolves, Meta’s strategy with Llama 2 demonstrates the challenges of balancing openness and monetization in the technology industry.
