Meta Enters Cloud AI Business, Ready to Challenge Amazon and Google

Meta is reportedly preparing a cloud business that sells access to AI computing and artificial intelligence models. This move will bring the Facebook parent company into direct competition with Amazon Web Services, Google Cloud, and Microsoft Azure. Illustration from Pexels photo

JAKARTA - After pouring in a giant investment for artificial intelligence, Meta is now preparing to open up new revenue sources. The Facebook parent company is said to be preparing a cloud business that sells AI computing and access to AI models to other parties.

TechCrunch, citing a Bloomberg report, said Meta is drawing up plans for a cloud infrastructure business. If it goes ahead, Meta will enter an arena that has been dominated by Amazon Web Services, Google Cloud, and Microsoft Azure.

The move comes after SpaceX, through xAI, took a similar path first. In early May, SpaceX signed an agreement with Anthropic to use all the computing capacity in the Colossus 1 data center. SpaceX then made a similar deal with Google and Reflection AI.

According to TechCrunch, quoted Thursday, July 2, this development signals that the winner of the AI race is not necessarily the company with the most advanced models and services. Those who control large data centers and computing capacity can also be in a key position.

However, these calculations only make sense if the demand for AI computing remains strong. A number of parties are beginning to worry that the development of AI infrastructure is too fast and could create a bubble. Moreover, expensive AI chips have a fast value. Other questions have also not disappeared. Are AI companies able to generate revenue from end users to cover bets worth trillions of dollars.

Meta doesn't seem too bothered by the doubts. By the end of the first quarter, Meta had committed to spending $182.9 billion on AI infrastructure in the coming years. The figure includes major projects in Louisiana and Ohio. The project in Ohio, which Mark Zuckerberg says will be the size of Manhattan, is expected to start operating this year.

Unlike Google and OpenAI, Meta has not seen significant demand for its AI models and services. Meta did not detail revenue from Meta AI or Llama in its financial statements. Llama is a family of open-weight AI models, which means that the models are more open to be used and developed by other parties than closed models.

Meta executives also emphasize more often the use of AI for the company's internal needs. This condition can show that Meta's AI business has not yet become a large independent source of revenue.

To pursue the returns from the jumbo spending, Meta is said to be able to imitate the CoreWeave business model. Bloomberg reported that the way is to sell access to raw computing capacity. This means that customers rent Meta's data center computing power to run their own AI work.

Bloomberg also reported that Meta was considering a move like AWS, namely selling access to various AI models hosted on Meta's infrastructure. One of them is Muse Spark, a closed-weight model that the company has just launched.

The new business line is reportedly under an initiative called Meta Compute. According to reports quoted by TechCrunch, this initiative is led by Meta Infrastructure Chief Santosh Janardhan, Meta Superintelligence Labs leader Daniel Gross, and President Dina Powell McCormick.

The report is in line with Zuckerberg's statement in May. At that time, he said Meta's cloud computing business "is clearly under consideration" as a way to earn returns on the company's massive investment in the development of superintelligence AI.