Skip to content 
Meta has been spending huge money on AI for a long time.Mark Zuckerberg just made that fact official. He made a new group called Meta Compute. This group will run the new data centers. It will handle the new chips. It will manage the energy deals. It will plan for future needs. This group reports directly to Mark Zuckerberg.
The size he is talking about is huge. He wants tens of gigawatts of power this decade. He wants hundreds of gigawatts later on. That massive scale is why this is a battle. It is not just a simple upgrade.
Meta’s Zuckerberg bets hundreds of billions on AI data centers
When you watch the video listen to how it tells the story. It frames this as a weapons race. In this race computing power equals control. This is not just about having more servers. This is Meta trying to get rare computer chips. It is trying to lock down long term power supplies. It is trying to get building permits faster. Meta does not want to get stuck renting computer power from someone else. It does not want to fall behind companies like Microsoft and Google and OpenAI. The video sets the mood. This is Mark Zuckerberg saying Meta will not be a loser. It will not come in last place.
People are reacting in different ways for a good reason. Some experts see Meta Compute as a message. It is proof to people with money and to workers that Meta is serious. It wants to lead in building AI systems. It does not want to just follow others.
Other people see a risky change. They think Meta is moving toward a model that costs a ton of money. This could hurt its profits. It could start to look more like a power company. It would not look like a software company that makes a lot of money on every sale.
A visual tour/breakdown of Meta’s Hyperion-scale data center build (the “Manhattan-sized” angle).
Metas $10B AI Data Center Hyperion is the Size of Manhattan
The bigger stake is what happens if Meta actually pulls this off: AI leadership stops being only about models and apps, and starts being about power contracts, financing, geopolitics, and build speed.That is why Meta Compute divides the work. One part handles the daily job of running their systems. The other part plans for their huge future needs. That is also why deals with the government are now on the company chart.
The second video makes the talk about gigawatts feel real. This is not just an idea anymore. This is about buying land and building with steel and finding enough electricity.
