Experts Sound the Alarm on the ‘Scary’ Energy Demands and Hidden Costs of the AI Boom
- Unprecedented Power Consumption: Sam Altman’s planned AI data centers could devour 10 to 17 gigawatts of electricity daily—equivalent to the combined peak demands of New York City and San Diego, or even entire countries like Switzerland and Portugal.
- Nuclear Ambitions vs. Harsh Realities: While Altman champions nuclear power as the future backbone for AI’s explosive growth, experts warn that building such capacity is years away, forcing reliance on renewables, natural gas, and other sources in the short term.
- Looming Environmental and Societal Toll: The rush for AI infrastructure risks derailing net-zero promises, straining water supplies, disrupting ecosystems, and generating toxic e-waste, sparking calls for urgent societal debate.
Imagine the glittering skyline of New York City on a blistering summer evening, where every skyscraper hums with air conditioners, subways rumble beneath the streets, and millions of lights pierce the night. Now layer on San Diego during a brutal heat wave, when the grid teeters on the edge of collapse under a surge past 5,000 megawatts. This isn’t just a vivid metaphor—it’s the staggering reality of the electricity that Sam Altman and his partners at OpenAI envision for their next generation of AI data centers. A single project could consume more power every day than these two major American cities at their breaking points, marking what experts call a “seminal moment” in the intersection of technology and energy.
Andrew Chien, a professor of computer science at the University of Chicago, has long anticipated this turning point, but its scale still alarms him. “Ten gigawatts is more than the peak power demand in Switzerland or Portugal,” he explained in a recent interview. “Seventeen gigawatts is like powering both countries together.” To put it in even starker terms, consider the Texas grid, where Altman recently broke ground on one of these ambitious projects. That grid typically handles around 80 gigawatts to support everything from oil refineries and factories to households across the state. Altman’s AI expansion? It could claim up to 20% of that total capacity— a “crazy large amount of power” that Chien says dwarfs other industries’ demands.
Altman himself frames this massive build-out as an unavoidable step forward in the AI revolution. “This is what it takes to deliver AI,” he declared during the Texas groundbreaking, pointing to the explosive growth of tools like ChatGPT, whose usage has skyrocketed tenfold in just 18 months. With demand showing no signs of slowing, Altman sees “compute infrastructure” as the foundation of tomorrow’s economy. His preferred energy source? Nuclear power. He’s invested heavily in both fission and fusion startups, arguing that only reactors can deliver the reliable, high-density energy needed to fuel AI’s insatiable appetite without the intermittency issues of renewables.
Yet, experts like Chien are quick to temper that optimism with cold facts. “As far as I know, the amount of nuclear power that could be brought on the grid before 2030 is less than a gigawatt,” he noted, highlighting a yawning gap between Altman’s 10 to 17 gigawatts vision and what’s realistically achievable. Nuclear, he added, represents “a ways off, and a slow ramp, even when you get there.” Instead, Chien predicts a mix of wind, solar, natural gas, and advanced storage technologies will fill the void in the near term, as the industry scrambles to keep pace.
Fengqi You, an energy-systems expert at Cornell University, offers a balanced but cautious view, suggesting nuclear might indeed become essential if AI’s expansion continues unchecked. However, he warns that “in the short term, there’s just not that much spare capacity”—regardless of whether it’s from fossil fuels, renewables, or nuclear. “A typical nuclear plant takes years to permit and build,” You explained, emphasizing that timelines for such projects often stretch into the double digits. For now, he believes companies like OpenAI will have to lean on quicker options: ramping up renewables, tapping into natural gas, and perhaps retrofitting older plants to squeeze out extra capacity. But even these stopgaps come with uncertainties—how to expand grids fast enough without triggering blackouts or skyrocketing costs?
The environmental implications of this AI-fueled energy binge are perhaps the most “scary” aspect, as Chien puts it. “We have to face the reality that companies promised they’d be clean and net zero, and in the face of AI growth, they probably can’t be,” he said. Beyond the obvious carbon emissions from fossil fuel backups, there are hidden tolls on ecosystems and communities. Data centers require enormous amounts of water for cooling—often in arid regions already grappling with scarcity—which could disrupt local biodiversity and strain resources. You echoed these concerns, noting that if these facilities “consume all the local water or disrupt biodiversity, that creates unintended consequences” that ripple far beyond the tech world.
Adding to the urgency is the rapid obsolescence of AI hardware. With Nvidia rolling out new processors like the Vera Rubin GPUs every year, old chips are discarded en masse, creating waste streams riddled with toxic chemicals. Chien calls for a “broader societal conversation” about these costs, urging accountability. “They told us these data centers were going to be clean and green,” he said. “But in the face of AI growth, I don’t think they can be. Now is the time to hold their feet to the fire.”
The financial stakes are equally mind-boggling, underscoring the broader economic shift Altman envisions. Each OpenAI site carries a price tag of roughly $50 billion, with total planned spending ballooning to $850 billion. Nvidia has committed up to $100 billion to support the effort, supplying millions of its cutting-edge GPUs. This isn’t just about building servers—it’s about reshaping global infrastructure, economies, and perhaps even geopolitics around AI’s needs.
Altman’s empire represents both the thrilling promise and perilous pitfalls of AI’s ascent. As data centers sprout across landscapes like Texas, the world must grapple with whether this power-hungry path is sustainable—or if it’s time to rethink the unchecked growth of artificial intelligence before it devours more than we can afford.