Artificial Intelligence (AI) is evolving fast—so fast that it’s not just testing the limits of technology, but also straining the world’s energy resources. Former Google CEO Eric Schmidt recently warned U.S. lawmakers that AI could, in the future, consume up to 99% of the world’s electricity. While the figure may sound extreme, his message is clear: the current trajectory of AI development is unsustainable without serious planning around energy. And he’s not alone—industry leaders like OpenAI’s Sam Altman have echoed similar concerns, revealing the massive energy demands behind today’s most powerful AI systems. As AI tools grow more advanced and widespread, powering them could become one of the biggest challenges of the digital age.
Former Google CEO Eric Schmidt recently made a bold and worrying prediction: artificial intelligence could eventually consume 99% of the world’s electricity. While that figure may seem far off, it’s based on a growing concern in the tech world—AI’s rapid advancement is outpacing the energy infrastructure needed to support it.
At present, AI uses around 3% of global electricity. But with more powerful models being developed and deployed at scale, this number is expected to increase dramatically. Schmidt told U.S. lawmakers that AI systems will need an additional 29 gigawatts of power by 2027 and 67 gigawatts by 2030—comparable to the output of dozens of nuclear power plants. The scale of growth is so massive that data centers powering AI could soon rival or surpass nuclear plants in energy use. Schmidt warned that this could place enormous stress on electricity grids, raise energy prices, and even lead to shortages if the expansion isn’t planned properly.
Adding urgency to this conversation, Schmidt highlighted that China is investing heavily in AI infrastructure. If countries like the U.S. don’t catch up in terms of computing power and electricity supply, they risk falling behind in the global AI race—something that could have deep economic and strategic consequences.
Supporting Schmidt’s concerns, OpenAI CEO Sam Altman also raised red flags about AI’s energy appetite. In a recent discussion, Altman said, “Our GPUs are melting,” referring to the strain that AI workloads are placing on computer hardware. He even joked that generating whimsical image prompts like “in Studio Ghibli style” were consuming so much power that they significantly increased costs. Even seemingly small and polite user habits, like typing “please” and “thank you” in prompts, can lead to millions of dollars in electricity bills when scaled globally.
Altman has also publicly acknowledged that future AI models will require breakthroughs not just in software, but in energy availability. Both Schmidt and Altman agree that we need to start preparing now—by building more efficient systems, expanding access to clean energy, and developing smarter ways to manage AI’s energy demands.
The AI revolution is unfolding quickly, but it’s not just a technological one. It’s increasingly an energy story too. If we don’t find a balance between innovation and sustainability, the very tools designed to make life better could end up draining the systems we all rely on.
