Surging artificial intelligence demand faces a "critical bottleneck" according to Goldman Sachs, with the fragile U.S. power grid failing to keep up with monstrous industry growth. In fact, China may have already won the battle for electric power dominance after decades of infrastructure investment, unlike our politically divided country.
AI data centers require an enormous amount of electric power to pump out a growing list of high-tech applications. These huge nests of servers, storage systems and networking equipment are gobbling up terawatts like candy, shifting U.S. GDP into the construction of new facilities. Consumer spending, which had controlled two-thirds of the nation's economic engine, is retreating.
Don't Miss:
- The same firms that backed Uber, Venmo and eBay are investing in this pre-IPO company disrupting a $1.8T market — and you can too at just $2.90/share.
- The ECG Hasn't Changed in 100 Years — This AI Upgrade Could Help Detect Heart Disease Years Earlier
Tech Buzz China founder Rui Ma says the Chinese government has already solved its energy needs by building enough power to drive massive data centers. "This is a stark contrast to the U.S.," she told Fortune magazine, where "AI growth is increasingly tied to debates over data center power consumption and grid limitations."
The price tag to fix this electricity gap will be enormous, with McKinsey forecasting $6.7 trillion in spending just to meet AI demand between 2025 and 2030.
Electricity expert David Fishman told Fortune that China adds more electric demand than the annual consumption of Germany each year but entire provinces are covered in rooftop solar, with a single province matching India's electricity supply. He says this historic build out is the result of central planners deliberately overbuilding and investing throughout the power sector."
Trending: Wealth Managers Charge 1% or More in AUM Fees — Range's AI Platform Does It All for a Flat Fee (and Could Save You $10,000+ Annually). Book Your Demo Today.
Goldman analysts Frank Long and George Lee hope that "smart demand management" will bridge the gap between U.S. electrical supply and demand until government and private industry play catch-up with trillions in new spending for data and power generation facilities. It says AI workloads can stop operations and leverage containment more efficiently than traditional cloud workloads, if control algorithms are properly built.
That code would execute a "fundamental shift" in AI workloads, adding "curtailment programs" to automatically stop operations and load balance when the U.S. power grid is under stress, freeing kilowatts for other uses. By dialing back for a few hours at a time, they can increase efficiency in training and inference tasks, applying greater power during the grid's slack periods.
See Also: They Sold Their Last Real Estate Company for Nearly $1B — Now They're Building the Future of U.S. Industrial Growth
Complicating AI growth, tech and utility industries aren't designed for the financial challenges required to meet future energy demand. As Fishman points out, large-scale infrastructure projects rely heavily on private investment, in which stakeholders often expect to profit within three to five years. But power projects can take up to a decade to build and generate income.
The public is now realizing electric utilities want them to pay for at least part of increased data center demand, and it’s not sitting well. According to Fortune, the average electricity bill in Ohio, for example, has jumped at least $15 per month this summer due to increased demand. Political action has been surging at the same time, with rallies and protests popping up across the country.
Read Next: Bill Gates Says Climate Change ‘Needs to Be Solved' — This Award-Winning Building Material Is Tackling It Head-On
Image: Shutterstock
