OpenAI CEO Sam Altman launched a spirited – some might argue unhinged – defense of the artificial intelligence (AI) industry’s resource consumption, dismissing viral claims about ChatGPT’s water usage as “insane” while arguing that AI’s energy efficiency should be measured against the decades of “training” required to create a human being.
Speaking on the sidelines of the India AI Impact summit in an interview with The Indian Express, Altman addressed the mounting scrutiny surrounding the environmental footprint of massive data centers. While critics have highlighted the gallons of water required to cool the servers powering Large Language Models (LLMs), Altman didn’t mince words.
“Water is totally fake,” Altman said, describing claims that single queries consume significant amounts of water as “completely untrue” and having “no connection to reality.” He noted that modern data centers are increasingly moving toward closed-loop liquid cooling systems rather than traditional evaporative methods that deplete local supplies.
The most provocative moment of the summit came when Altman refuted comparisons, including those previously suggested by Microsoft Corp. co-founder Bill Gates, regarding the efficiency of the human brain versus AI.
Altman argued that critics often unfairly compare the massive energy cost of training an AI model to a single human action, ignoring the immense biological “overhead” of human development.
“It takes like 20 years of life, and all the food you eat before that time, before you get smart,” Altman said. “The fair comparison is: once a model is trained, how much energy does it take to answer a question versus a human? Measured that way, AI has probably already caught up.”
As governments scramble to balance AI leadership with net-zero climate goals, Altman’s comments signal an industry shift from apologies and toward a narrative that frames AI as an increasingly efficient, and perhaps inevitable, evolution of intelligence.
The comparison to human biology sparked immediate pushbacks from other tech leaders. Sridhar Vembu, co-founder of Zoho Corp., criticized the rhetoric on X. “I do not want to see a world where we equate a piece of technology to a human being,” Vembu said.
Despite his dismissal of water concerns, Altman conceded total energy consumption is a “fair” concern as AI integration scales globally. He emphasized that the industry’s growth is contingent on a rapid transition to high-output energy sources.
“We need to move towards nuclear or wind and solar very quickly,” Altman said, echoing a growing sentiment among tech giants who are currently outstripping the capacity of local power grids.
The scale of the challenge is significant. According to the International Monetary Fund, global data center electricity consumption in 2023 was already comparable to the national usage of France or Germany. The surge has led to local friction; recently, officials in San Marcos, Texas, blocked a $1.5 billion data center project following intense public opposition over grid stability and rising utility costs.
Beyond the environmental debate, Altman touched on industry flashpoints. He downplayed fears of AI-driven unemployment, suggesting the technology would create new categories of work. He also dismissed the idea of space-based data centers as “unlikely” this decade because of launch costs and hardware fragility. And he admitted that not taking equity in OpenAI was “a dumb thing” and suggested a reconciliation with Elon Musk was less likely than ending the global monopoly on chip manufacturing.
“The real conversation about Sam Altman’s comments isn’t per-query efficiency: it’s whether AI infrastructure build-out is outpacing clean energy capacity at the grid level,” said Mitch Ashley, vice president and practice lead, Software Lifecycle Engineering, at The Futurum Group. “And, whether data center siting decisions are creating localized resource pressure in markets like India where that tension is sharpest.”
“The analogy reveals a more consequential pattern,” Ashley added. “When AI leaders reframe environmental critique as a measurement dispute, they shift accountability away from infrastructure planning decisions and toward critics. Enterprises evaluating AI at scale need energy and water footprint transparency that holds across cooling architectures, not analogies that collapse under scrutiny.”
