### How Solar Is Powering the AI Revolution with High-Solar Microgrids
The rapid growth of artificial intelligence (AI) has brought new challenges to the energy sector, especially as datacenters used for training AI models demand increasing amounts of power. With estimates placing AI training datacenter power demand at 15-150 GW by 2030, the solar industry stands poised to play a pivotal role in meeting these energy needs sustainably. Let’s break down how solar companies and microgrids are shaping the future of AI energy infrastructure.
—
### Why AI Training Datacenters Need Solar
Artificial intelligence training is incredibly energy-intensive. Unlike inference datacenters—which process live user queries and need to be located close to consumers—training datacenters can be situated in remote areas since they only develop and fine-tune AI models. This geographic flexibility opens the door for innovative energy solutions, such as off-grid solar microgrids, particularly in regions with high solar potential like the southwestern United States.
What makes solar particularly viable for these facilities is the abundance of inexpensive land bathed in sunlight across the Southwest. Researchers have identified land parcels with the potential to host a staggering 1,200 GW of off-grid solar-plus-storage capacity equipped with natural gas
Original Article: https://pv-magazine-usa.com/2024/12/20/off-grid-solar-plus-storage-could-power-data-centers-used-for-training-ai-models/
