To one blog Last July, Meta Mark Zuckerberg CEO said “Access Sale” on Meta’s Llama AI models ” [Meta’s] Business model. “However, Meta does at least some Money from Llama through revenue distribution agreements, according to a newly acquired court registration.
THE depositsubmitted by lawyers for the plaintiffs in copyright lawsuit Kadrey v. Meta, in which Meta is accused of drawing up Llama models in hundreds of terabytes of pirate e -books, reveals that Meta “shares a percentage of revenue” that companies hosting the Llama models created by users of these models.
The deposit does not indicate which specific hosts pay meta. But Meta lists a number of reception partners in various blog seatsIncluding AWS, Nvidia, Databricks, Groq, Dell, Azure, Google Cloud and Snowflake.
Developers are not required to use a Llama model through a host partner. Models can be uploaded, refined and operating in a series of different material. But many hosts provide additional services and tools that make Llama models and work simpler and easier.
Zuckerberg reported licensing access to Llama models During a profit call last AprilWhen he also put in revenue from Llama in other ways, such as through business and advertising messaging services in “AI interactions”. But does not describe details.
“[I]F You are someone like Microsoft or Amazon or Google and you will basically resell these services, this is something we think we have to get some of the revenue, “Zuckerberg said.
More recently, Zuckerberg claimant That most of the Meta value comes from Llama comes in the form of improvements in models from the AI research community. Meta uses Llama models to supply a range of products on all its platforms and properties, including Meta AI assistant, Meta AI.
“I think it’s a good job for us to do this in an open way”, Zuckerberg Said during the profits of Meta Q3 earnings 2024. “[I]T makes our products better and not if we were only on an island building a model that no one was somewhat standardized around the industry. ”
The fact that Meta can produce revenue in a rather direct way from the Lama is important because the plaintiffs at Kadrey by Meta claim that Meta not only used pirate projects to develop Llama, but facilitated the violation of “sowing” or uploading these projects. The plaintiffs claim that Meta used methods of disguised methods to obtain e-books for training and in the process of ways in which torrenting projects were shared by e-books with other torrenters.
Meta plans to significantly increase its capital spending this year, largely thanks to its increasing investment in AI. In January, the company said it would spend $ 60 billion-80 billion on Capex in 2025-about the Capex of Doubout Meta in 2024-mainly in the Database and the Development Group of the AI Groups.
It is likely to compensate for part of the cost, Meta says it is considering starting a Meta AI subscription service that will add unspecified possibilities to the assistant.
Updated 3/21 at 1:54 pm: A Meta spokesman pointed TechCrunch to This profit call transcription for additional framework. We have added a Zuckerberg excerpt from it – specifically an excerpt about Meta’s intention to the share of the Revenue with large hosts of the Lama models.
