By Ryan Hinske
Medill Reports
SAN FRANCISCO — The Bay Area was the site of the biggest media event of the calendar year Sunday. At Super Bowl LX, Levi’s Stadium needed an amount of energy that would power thousands of homes. But across town, data centers make the big game look microscopic.
The stadium’s instantaneous demand surged to an estimated 12 megawatts by kickoff, based on energy demands from Super Bowl LIX in New Orleans. The scale of this demand is staggering until compared with some of the largest data centers that are increasing their power intake due to artificial intelligence. Incorporating AI servers into data centers requires hardware that uses stronger cooling systems and more physical space. According to a September 2025 Cornell study, larger facilities can demand 100 MW or more.
That means powering a large AI data center would require more than eight Super Bowls’ worth of energy production.
Silicon Valley is one of the most densely concentrated data center markets in the world, with the city of Santa Clara serving as its primary heartbeat. According to the History Cooperative, beginning in the 1930s, the area near Stanford University became a place for tech entrepreneurs to build companies without having to move east. Semiconductor manufacturing was the first innovation to boom in Silicon Valley before anywhere else, bringing companies such as Intel to establish research centers in the area. Personal computers followed in the 1970s, then the Dot-Com bubble as the internet took hold of the tech universe. Now, dozens of massive tech companies recognized by the average American — such as Google, Adobe and Salesforce — are based in the Bay Area.
The high concentration of data centers is largely powered by Santa Clara’s municipal utility, Silicon Valley Power, which says it offers significantly lower electricity rates than neighboring regions on its website. SVP also powers Levi’s Stadium. At the end of 2024, Silicon Valley had a capacity for 810 MW of energy production, according to a report by commercial real estate company Jones Lang Lasalle.
Still, the industry is primed to surpass SVP’s and the broader region’s capabilities.
Kazunori Okada is a computer science professor at San Francisco State University and studies intelligent computing.
“(AI) is on an exponential growth,” Okada said. “Personally, the main issue I think it has is that such growth is driven by unregulated corporate actions on maximizing their profit.”
Okada said there are many small language models in development, but these models cannot match the “increase of power consumption as more and more people use it for more tasks in our lives.” Therefore, Silicon Valley’s being a place where small tech companies can develop new and innovative AI language models is at risk.
OpenAI said it has plans for a 520 MW “gigafactory,” Stargate Norway, in Kvandal, northern Norway, which will host 100,000 NVIDIA graphics processing units. Norway is an attractive place for OpenAI because of abundant resources and a lack of facilities that require similar power requirements to that of Silicon Valley facilities.
“Globalization is dying,” Okada said. “The Bay Area’s reputation as the birthplace of AI technology could not last forever. … The Bay Area has enjoyed access to top brains. That has been the leverage. The way education is targeted by immigration laws is killing this advantage today. Many top brains come to the U.S. from other countries to study here. They’ve stopped coming here today.”
The Stanford study shows Northern California is constructing another 1,394 MW of capacity, creating a 172% increase. Other regions of the country are primed to outpace Silicon Valley: Las Vegas/Reno, Phoenix, Chicago, Dallas-Fort Worth, Atlanta and Northern Virginia, a region that is predicted to reach a whopping 11,077 MW capacity.
The Stargate Norway project is one of the first looks into how much energy AI data centers may need to produce in the 2030s. Stargate Norway would require more than 43 Super Bowls’ worth of energy to operate.
Investment in AI technology is booming. According to Reuters, NVIDIA’s valuation — the total estimated price tag for a business’s entire worth — went from $302 billion in 2022 to $4.3 trillion in 2025. But massive electricity projects are demonstrating the required investment into maintaining future AI capabilities.
In an email to Medill Reports, NVIDIA detailed its emphasis on energy efficiency in the 2025 Climate Week conference in New York City and its Rubin platform, which is in development and contains similar green initiatives for sustained intelligence production.
Ryan Hinske is a sports media specialization graduate student at Medill.