Meta plans a cloud business to sell spare AI computing power

In short: Meta is reportedly planning to sell access to its extra AI computing capacity and some of its AI models through a new cloud business.

What happened

Bloomberg reported that Meta is developing plans for a cloud infrastructure business. In simple terms, that means Meta would rent out its computing power, the same way some companies rent out storage space.

The product would sell “AI compute,” which is the processing power needed to run and train AI systems. You can think of it like renting time on very powerful computers inside large buildings called data centers (warehouses full of servers).

The report says this effort is part of an initiative reportedly called Meta Compute. It is led by Meta infrastructure head Santosh Janardhan, Meta Superintelligence Labs leader Daniel Gross, and Meta president Dina Powell McCormick.

Meta has spent heavily on this kind of infrastructure. TechCrunch notes Meta had committed to $182.9 billion in spending on AI infrastructure in coming years, and it has major data center projects in Louisiana and Ohio.

Meta’s move would put it in more direct competition with the biggest cloud providers, including Amazon Web Services, Google Cloud, and Microsoft Azure. Meta is also considering selling access to multiple AI models hosted on its systems, including its recently launched closed-weight model Muse Spark (a model where the underlying “settings” are not shared publicly).

Why it matters

For people who do not work in tech, this matters because the companies that own the “picks and shovels” of AI, meaning the data centers and chips, may make money even if their own AI apps are not the most popular. It could also affect prices and availability for businesses that rely on rented computing power to build AI features.

Source: TechCrunch AI