Meta CEO Mark Zuckerberg today (January 13th) via ThreadsAnnounceThe US will launch a massive project called "Meta Compute." This project has a rather aggressive goal: to build a gigawatt-scale AI infrastructure within the next decade, with a long-term target of hundreds of gigawatts, attempting to accelerate the realization of super intelligence through a "national-level" deployment of energy and computing power.
The computing power scale is comparable to that of a national-level power consumption.
In his post, Mark Zuckerberg revealed that Meta plans to build computing facilities with a scale of tens of gigabytes over the next decade.
What does this mean? 1 GW of electricity is roughly enough to power the daily electricity needs of 75 average American households. If Meta's long-term goal is "hundreds of GW," it means that the power consumption of its AI network will be equivalent to the total electricity consumption of a medium-sized country. This also highlights that Meta's estimated computing power requirements for future AI models (possibly Llama 5, Llama 6, or even later versions) far exceed current expectations.
The Golden Triangle: Engineering, Strategy, and Diplomacy Working in Tandem
To execute this massive "Meta Compute" project, Meta has also established a clear leadership structure:
• Commander-in-Chief:Santosh Janardhan, Global Head of Engineering, will be responsible for overseeing the implementation of the technology across the entire infrastructure.
• Long-term strategy:Daniel Gross, the former CEO of Safe Superintelligence who recently joined the team, will lead a new group focused on long-term capacity strategy, supplier negotiation, and business modeling.
• Diplomacy and Finance:Newly appointedNew President Dina Powell McCormickShe will play a crucial role. Leveraging her background in the White House and Goldman Sachs, she will be responsible for liaising with governments and sovereign entities to resolve the most challenging infrastructure construction permits, regulatory hurdles, and financing issues.
Embracing Nuclear Energy: Signing a 20-Year Power Purchase Agreement
But where does the electricity for such massive computing power come from? With unstable green energy supply and limited fossil fuels, nuclear energy has become the core solution for Meta.
As a key part of its plan, Meta announced a 20-year power purchase agreement with energy giant Vistra, securing power supplies for three nuclear power plants in the central United States. Furthermore, Meta confirmed collaborations with two companies developing small modular reactors (SMRs), highlighting that stable and clean nuclear energy has become the only solution to support its AI ambitions amid surging U.S. electricity demand driven by big tech companies.
