Meta estimated that its generative AI product line could generate between $460 billion and $1.4 trillion in revenue by 2035, according to recently unsealed court documents. The documents, revealed as part of an ongoing lawsuit filed by several authors against Meta, outline revenue forecasts made internally by the company.
Meta’s projections detail that its generative AI technologies would initially produce between $2 billion and $3 billion in revenue by 2025. By the following decade, revenue from these products is expected to grow rapidly, potentially breaking the trillion-dollar mark by 2035. However, the disclosure did not define specifically what Meta considers within its generative AI product bracket.
It is widely known that Meta already profits from and plans further monetization of its generative AI offerings. Some avenues include revenue-sharing agreements based on the utilization of its popular, openly available model family, Llama. Meta also recently initiated an API service that allows businesses and developers to customize these AI language models.
Furthermore, Mark Zuckerberg indicated that Meta’s AI-powered assistant, branded as Meta AI, may soon be monetized through advertisements and a paid-tier subscription, a strategy outlined in the company’s latest quarterly earnings call.
Achieving these ambitious revenue targets involves substantial investment. Documents reveal that Meta’s spending on generative AI groups reached over $900 million in 2024 and is expected to surpass $1 billion this year. These expenditures do not include the significant investments required in infrastructure, data centers, and hardware necessary to sustain and enhance Meta’s artificial intelligence capabilities.
Earlier statements from Meta noted that its 2025 capital expenditures would range between $60 billion and $80 billion, chiefly dedicated to the construction and expansion of new data centers. Interestingly, financial records also showed that Meta had once considered licensing agreements with authors — potentially investing up to $200 million for training content alone, half of which would directly target book rights acquisitions. The company, however, reportedly abandoned this licensing plan and instead faces allegations of large-scale unauthorized use and distribution of authors’ works to train its AI models.
At present, Meta has not publicly responded to questions or commented directly on these newly uncovered figures.
