Mark Zuckerberg had big ambitions for Meta's AI business. Billions of dollars have been poured into this initiative, top-tier AI researchers have been hired, talent has been aggressively poached from competitors, and the company has spent years trying to catch up in the generative AI race.
Yet, despite all that investment, Meta is still struggling to establish itself alongside industry leaders like Anthropic or OpenAI. Now, Facebook's parent company appears to be seeking another way to monetise its AI ambitions by renting out its artificial intelligence infrastructure.
According to a Bloomberg report, Meta is considering turning its AI infrastructure into a cloud business by allowing other companies to rent its computing power-even as its own flagship AI model faces delays and remains unavailable to developers.
The report notes that Meta is evaluating two options. One would allow developers to access AI models hosted on Meta's infrastructure and pay only for the computing power used, similar to Amazon Web Services' Bedrock platform. The other option would involve Meta renting out its surplus AI computing capacity, thereby competing directly with AI-specialised cloud providers like CoreWeave, as well as cloud giants such as Amazon Web Services, Microsoft Azure, and Google Cloud.
The aforementioned cloud business is expected to operate under an internal project dubbed "Meta Compute," which is tasked with managing the company's growing AI infrastructure.
According to the report, the strategy is still under discussion and could change. Meta has not commented on the matter.
But why does Meta want to sell cloud services and computing capacity?
Meta's cloud plans come at a time when major tech companies are facing growing pressure to monetize their massive investments in AI. Since OpenAI's ChatGPT sparked the generative AI boom in late 2022, companies have poured billions of dollars into data centers and Nvidia graphics processing units (GPUs) to develop and run AI models.
In fact, Meta is one of the biggest investors in AI. Earlier this year, the company stated it could invest up to $145 billion in AI infrastructure by 2026. Zuckerberg has also indicated that Meta's total AI investment could reach around $600 billion through 2028, adding on Threads that the company might spend even more if AI continues to advance rapidly.
After investing hundreds of billions of dollars in AI, Meta now appears to be looking for ways to turn that investment into a new business.
Meta faces challenges in the AI race
It is worth noting that Meta's plans to make money by selling AI computing power come as the company is still trying to catch up with rivals in the AI race. While companies like OpenAI, Anthropic, and Google have launched increasingly powerful AI models, Meta's own AI roadmap has faced delays.
In April, Meta unveiled Muse Spark, the first AI model from its revamped AI team. However, Reuters reported that the model has not yet been made available to developers, while a Wall Street Journal report indicated that there is currently no launch date. This delay has raised questions about whether Meta can catch up to its rivals despite having invested billions of dollars in AI talent and infrastructure.
Nevertheless, Zuckerberg has hinted that selling AI computing power has been on the table for some time. During Meta's shareholder meeting in May, it was stated that companies approach Meta "almost every week" to request access to its AI models or surplus computing capacity; it was further noted that entering the cloud computing business was an option "definitely on the table."
For their part, investors appeared to welcome the idea. Meta shares rose as much as 10% during Wednesday's trading session before closing up nearly 9%, as investors bet that the cloud business could help generate a new revenue stream from the company's massive investments.