"NERC issued the Level 3 alert for "essential action" — the agency's highest level — after issuing two previous warnings about data centers in the last nine months. The agency sends alerts to the entities that make the electric grid run smoothly, from regional transmission organizations to power plant owners. This time, the agency stressed that grid operators lack "sufficient processes, procedures, or methods to address risks associated with computational loads.""
"NERC says that data centers — particularly ones running AI workloads — are prone to wild power swings. The constant, rapid fluctuations between extremely high and extremely low power use levels can put entire electric grids at risk of going offline. Power swings "can occur in seconds, leaving 'little or no room for real-time responses,'" the report said."
"It's not only AI training causing the instability. In its report, NERC said crypto mining and traditional data centers threaten the grid. NERC asked grid operators to take immediate action and file risk mitigation plans with the agency by August 3. "As the grid faces unprecedented challenges from a surge in large power consumers, NERC is taking significant steps to ensure the reliability of the bulk power system," the agency said in a press release on Monday."
NERC issued a Level 3 alert urging electric grid operators to take essential action to address new risks from rapidly growing data center demand. Grid operators lack sufficient processes, procedures, or methods to manage risks tied to computational loads. NERC warns that data centers can cause wild power swings, with rapid fluctuations between extremely high and extremely low power use levels. These swings can occur in seconds, leaving little or no room for real-time responses and increasing the risk of parts of the grid going offline. The risk is not limited to AI training; crypto mining and traditional data centers also threaten grid stability. Operators must submit risk mitigation plans by August 3.
Read at www.businessinsider.com
Unable to calculate read time
Collection
[
|
...
]