Businesses continue to invest in new infrastructure to support artificial intelligence, sending the demand for power soaring in data centers. With data center operators rethinking their approaches to power management in the emerging era of AI, downtime is becoming more costly than ever, especially amid increasingly volatile weather events.
Experts are forecasting a particularly intense hurricane season this year, bringing the distinct possibility of power outages. Learning from the major power outage events from 2023 can help guide the way forward for operators seeking to enhance their power management strategies and bolster disaster preparedness.
Six Key Blackouts from 2023:
- Wild weather travel cross-country – On February 22, few states were left unaffected by extreme weather. While a winter storm closed interstate highways from Arizona to Wyoming and prompted the first blizzard warning in Southern California in decades, record highs were broken throughout cities in the Midwest, mid-Atlantic, and Southeast. The events caused significant outages, with more than 811,000 recorded in Michigan; 58,000 in Wisconsin; and over 41,000 in California, among others.
- A crawling, catastrophic storm – Nearly one million customers were left in the dark after a powerful winter storm made its way across the Great Lakes, South, and mid-Atlantic states on March 3, causing crews to work around the clock to restore power. The next day, hundreds of thousands remained without power in Kentucky, Tennessee, and Michigan, and even more in Alabama, West Virginia, Indiana, Ohio, Pennsylvania, and New York.
- Torrential rain washed out power – More than 500,000 homes and businesses were without power across California on January 8 as the state faced heavy rain, high winds, and potential floods. Areas across the state prepared to endure five to seven inches of rain over a nine-day period.
- Midwestern mess – On June 29, a line of severe thunderstorms produced rain, hail, and winds up to 70 mph, leaving nearly 500,000 people facing power outages in Indiana and Illinois. Utilities warned customers to expect more outages as forecasts called for “an unsettled weather pattern” to move into the area.
- Biggest outage in three decades levels San Francisco – The San Francisco Bay Area suffered its worst power outage since 1995 on March 14 due to a week-long storm. The heavy rains and high winds left 450,000 customers in the dark. In the Santa Cruz Mountains, wind gusts hit 97 mph as downed trees toppled power lines.
- Floods, winds and widespread blackouts – A powerful storm in New England brought over three inches of rain on December 19, leading to floods and power outages for over 600,000 people. The storm’s aftermath saw thousands of workers rushing to restore power amidst extensive damage.
As data centers are the key to powering innovations like AI, these examples underscore the importance of comprehensive disaster preparedness in the face of unexpected events. Luckily, data center operators can leverage a wide array of technologies and solutions to help keep the power on and protect critical assets amid the AI boom.
Power management tools boost resiliency
To guard against disruptions during the summer storm season, data center operators should evaluate existing power management resources and seek out opportunities to upgrade equipment where possible. Innovations in both hardware and software have advanced the ability to meet today’s power management challenges, helping current and future data centers effectively maintain uptime.
As an example, technologies like uninterruptible power supplies (UPSs), which serve as a bridge to generator power during an outage, have evolved to deliver high power density in a compact footprint, ensuring optimal space for additional equipment. Lithium-ion batteries, which extend the lifespan of UPSs, are critical components of a robust disaster preparedness strategy. Operators should determine whether their vendors can offer modernization services to extend the life of existing systems through product refurbishment, a cost-effective approach to enhancing power management without requiring wholesale replacement of existing infrastructure.
In addition to hardware devices, digital solutions like data center performance management software (DCPM) and distributed IT performance management software can help improve oversight of critical assets, both on premise and offsite through integration with network management cards. Using these capabilities, operators can manage and monitor assets to better understand equipment health and gracefully shut down equipment in the event of power anomalies. Adding services that use predictive analytics can enable operators to track and analyze power trends to make better, more informed decisions.
Finally, a well-rounded disaster preparedness strategy should also include industrial-grade hardware tools such as surge protectors, power strips, and rack enclosures. Combining power management solutions with rackmount and cabinet power distribution units (PDUs) work to keep valuable assets safe during threatening events.
Ready for whatever comes
As AI continues to evolve, more companies will look to invest in power equipment that supports the deployment of GPUs and other advanced technologies. By developing a comprehensive backup power strategy that integrates innovative hardware and software solutions, data center operators can feel secure in their ability to protect critical assets and guard against whatever Mother Nature can dish out.
