VIEWPOINT: AI’s Explosive Energy Demand Raises Alarms
‘GENIUS ARCHIVES’ POINTS TO accelerating pollution, automation, and erasure in Black neighborhoods across America
By Brie Alexander
August 2025 – The cost of AI’s energy appetite is staggering, with a 2025 study by the Energy Justice Network indicating that 9 of the 10 counties most impacted by new AI data center projects are poor, predominantly Black, or rural communities already suffering from environmental burdens.
The series, focused on Pan African affairs, shares that training a large AI model can consume more electricity than 100 U.S. homes use in a year (International Energy Agency, 2024). Inference, the process of running AI models, requires even more as usage scales.
Electricity consumption is expected to double by 2030, with AI being the leading cause. In the U.S., AI data centers are projected to account for nearly 50% of electricity demand growth over the next five years. By 2028, they may consume up to 12% of all U.S. electricity (IEA, 2024).
Alexander explains that an AI rack can draw 50–100kW, five times more than traditional servers, requiring massive cooling systems – often powered by methane gas turbines. In many areas, these backup plants operate 24/7, placed directly beside homes in poor and often Black neighborhoods.
Alexander in her series shares that modern AI data centers require minimal staffing so that a $10B campus might employ only 100–200 people, most of them highly skilled and recruited from elsewhere. Local Black residents are rarely hired for long-term roles.
Worse still, the very technology inside these centers is designed to automate the sectors where Black workers are overrepresented—from transportation and retail to basic healthcare and customer service. A report by the Brookings Institution (2023) warns that up to 44% of jobs held by Black Americans are at high risk of automation. Without meaningful policy shifts, these communities face economic displacement alongside environmental degradation.
“In the world they’re building, we are either deleted or disposable,” writes Alexander. “This isn’t science fiction. It’s source code.” By 2028, AI data centers could account for 12% of total U.S. electricity use and are largely powered by methane gas turbines running beside homes in underserved communities, she said.
Alexander explained that despite billion-dollar headlines, AI data centers typically hire fewer than 200 workers – most recruited from outside the community. Meanwhile, the technology housed inside is automating jobs that Black Americans disproportionately hold: transportation, customer service, retail, and basic healthcare.
A Brookings Institution report warns that up to 44% of jobs held by Black workers are at high risk of automation but policy safeguards remain minimal. “The lie is that tech jobs are coming,” says Alexander. “What’s really coming is job loss, displacement, and data-driven control.”
In stark contrast, China is developing ocean-cooled data centers to mitigate energy use and environmental impact. U.S. tech companies, by comparison, are placing facilities in historically Black communities, relying on low regulation and high vulnerability, said Alexander.
Brie Alexander is a Pan-African cultural strategist, and founder of the Genius Archives initiative, a new series tracing global footprints, from cobalt mines to cloud servers, from vanishing villages to AI campuses.
Ends