Meta Plans AI Cloud Service to Challenge Tech Giants

Meta Platforms is developing a cloud infrastructure business to sell AI computing power and model access, positioning itself as a direct competitor to Amazon Web Services (AWS), Microsoft Azure, and Google Cloud. The initiative is part of an internal effort called Meta Compute, which aims to monetize the company's significant investment in data centers and chips. By offering access to its own models, such as the Muse Spark, or providing raw computing capacity, Meta intends to generate revenue from excess infrastructure as it continues its push toward building 'superintelligence.'

The move comes after Meta committed hundreds of billions of dollars to data center capacity and specialized chips. While investors have previously expressed concern over the scale of this capital investment, the launch of a cloud business offers a viable pathway to return value from these massive infrastructure projects. Meta is currently considering multiple models for this business, including an API-based service measured in 'tokens' for AI usage, and a neocloud strategy similar to competitors like CoreWeave Inc. Meta President Dina Powell McCormick and other senior leaders are spearheading this initiative.

The demand for AI computing remains high, and Meta’s entry could reshape the competitive landscape for cloud services. Chief Executive Mark Zuckerberg has previously indicated to shareholders that the company is open to selling computing power if it deems the capacity overbuilt. This strategy is part of a wider trend in the tech industry to capitalize on the AI boom through infrastructure rentals. For India, a major hub for AI startups and developers, the availability of Meta’s cloud infrastructure could provide new avenues for innovation and model training, potentially lowering the barriers to entry for local firms relying on existing US-based cloud providers.