Sam Altman Defends AI Energy Use by Comparing It to Human Metabolism

OpenAI CEO Sam Altman has ignited a global conversation about the environmental cost of artificial intelligence by drawing a provocative parallel between the energy demands of training AI systems and the biological energy required to raise and educate a human being. During a January 2023 interview with tech journalist Connie Loizos, Altman argued that while AI data centers consume significant electricity, the human developmental process — from infancy through higher education — requires vastly more energy over a lifetime. "It takes a lot of energy to train a human too," Altman reportedly said, suggesting that societal focus on AI’s carbon footprint may be disproportionate when compared to the immense resources invested in human capital.

According to Financial Express, Altman’s remarks were part of a broader discussion on AI safety and scalability, where he emphasized that the energy used to train large language models like GPT-4, while substantial, is a one-time or infrequent investment compared to the continuous, lifelong energy consumption of a human being. He cited metabolic rates, food production, transportation, housing, and education systems as components of human energy expenditure that collectively dwarf the power needs of even the largest AI training clusters.

However, Altman’s analogy has drawn sharp pushback. Sridhar Vembu, CEO of Zoho Corporation, countered in a public response reported by India Today and MSNBC, that while the comparison may be theoretically interesting, it risks obscuring the urgent, scalable environmental impact of AI infrastructure. "We’re not training one human; we’re deploying millions of AI models globally, 24/7, with rapidly growing demand," Vembu stated. "The energy intensity per inference is orders of magnitude higher than human cognition, and the scale is exponential."

Indeed, recent studies from the University of Massachusetts Amherst and the International Energy Agency (IEA) estimate that training a single large AI model can emit as much carbon as five cars over their lifetimes. Data centers, which power AI services, already consume roughly 1% of global electricity — a figure projected to triple by 2030. Altman’s argument, while philosophically compelling, does not negate the need for energy-efficient hardware, renewable-powered infrastructure, or regulatory oversight.

Environmental advocates caution against using human energy consumption as a moral shield for AI’s ecological burden. "Comparing apples to oranges doesn’t reduce emissions," said Dr. Lena Ruiz, a sustainability researcher at Stanford. "We need to optimize AI’s footprint, not justify it by invoking human biology."

Altman has since acknowledged the complexity of the issue, noting in later interviews that OpenAI is investing in renewable energy partnerships and exploring more efficient model architectures. Yet the core tension remains: How do we balance innovation with planetary responsibility? The debate transcends technical metrics — it’s a question of values, priorities, and who bears the cost of progress.

As AI becomes increasingly embedded in daily life — from healthcare diagnostics to content creation — the energy debate will only intensify. While Altman’s analogy may offer perspective, it cannot substitute for accountability. The future of AI must not be measured solely by its intelligence, but by its sustainability.