AI's Energy Revolution: The Gobi Project
New ways to efficiently use energy in the age of artificial intelligence.
1AI's Energy Problem
The biggest barrier in AI development is no longer chips but energy consumption. According to Goldman reports, U.S. data centers' energy demand will reach 31 GW by 2025. This means AI is competing with society for energy. Power grids are increasingly struggling to meet this demand.
2Mission Gobi
Envision CEO Lei Zhang unveiled the Gobi Project at VivaTech, aiming to create 5 GW of green energy capacity in deserts by 2030. This project seeks to prevent AI from burdening urban grids by building renewable energy systems specifically designed for AI.
3Energy and AI Integration
Envision's system consists of three layers: an intelligent operations hub, Physical AI supported by the Tianji Weather Model and Dubhe Energy Model, and advanced energy infrastructure. This integration balances energy generation, storage, and AI processing in real time.
4Inner Mongolia Example
Envision has established a 2 GW system in Chifeng, Inner Mongolia, powered entirely by renewable energy. This system integrates wind, solar, storage, and hydrogen to provide a sustainable energy model for AI. The gigawatt-scale campus in Ulanqab is an example of this model.
5Space and Desert Solutions
While SpaceX plans data centers in space powered by uninterrupted solar energy, Envision focuses on deserts. Both approaches aim to use the most efficient resources to meet AI's energy needs. The common point is that AI should not compete with humanity for energy.
6Energy Model for the AI Era
The industrial age was shaped by coal, the electrical age by grids. The AI age could be built on energy abundance. The Gobi Project aims to position AI infrastructure near energy sources to offer a sustainable future. This could be a revolution that reduces AI's societal cost.
Original article: www.theregister.com - Articles — AI'nın Enerji Devrimi: Gobi Projesi