Key Points:
- Sam Altman called rumors about ChatGPT’s massive water use completely fake.
- He noted that modern data centers no longer use water for evaporative cooling.
- Altman admitted the total energy needed to run AI systems is a valid concern.
- The CEO argued that answering a question with AI takes less energy than a human brain.
OpenAI CEO Sam Altman strongly pushed back against rumors regarding the environmental cost of artificial intelligence. Speaking at an event hosted by The Indian Express during a major AI summit, Altman tackled the growing debate over how much water and power his company uses.
He directly addressed viral claims that ChatGPT drinks massive amounts of local water. Altman called these rumors totally fake and insane. He pointed out that people online often say each user query uses up to 17 gallons of water. He firmly stated that these numbers have absolutely no connection to reality.
Altman explained exactly where the confusion started. He acknowledged that water use was a real issue in the past when tech companies relied heavily on evaporative cooling systems for their server farms. However, he clarified that they no longer use that outdated cooling method today.
While he dismissed the water myths, Altman admitted that power consumption remains a genuine problem. Because the world uses so much AI right now, the total energy footprint is huge. He told the audience that people are right to worry about this massive electricity demand.
To solve this energy problem, Altman called for a quick shift in how we generate power. He urged global leaders to move away from fossil fuels and build more nuclear, wind, and solar energy plants. He also denied claims that a single ChatGPT prompt burns through the equivalent of one and a half iPhone batteries, saying the real number is nowhere near that high.
Altman then offered a unique way to look at energy efficiency. He complained that critics unfairly compare the energy needed to train a massive AI model from scratch against a human doing a simple task. He reminded everyone that training a human takes about twenty years of food, resources, and learning just to get smart.
He argued that a fairer test compares a fully trained AI against a human answering the exact same question. Measured that way, Altman believes artificial intelligence has already beaten human energy efficiency.
