Sam Altmans Absurd Defense of AI Energy Crisis

When your best argument is humans also use energy, you know you are in trouble.

Sam Altman has a new favorite talking point: sure, AI uses a lot of energy—but so do humans!

This is not a joke. This is actually what the CEO of OpenAI said at a recent event in India when confronted with concerns about AI environmental impact.

The Water is Fake Claim

First, Altman dismissed concerns about AI water usage as totally fake. He claimed the switch from evaporative cooling means AI data centers no longer have water issues.

This is misleading at best.

While some data centers have moved away from evaporative cooling, the water problem extends far beyond that. Training large language models requires enormous amounts of water for cooling, and inference every time you type something into ChatGPT also consumes water.

Researchers have documented significant water usage at AI facilities. The claim that the issue is completely untrue ignores independent scientific studies. It is corporate PR masquerading as environmental expertise.

The Humans Use Energy Too Deflection

But the real gem is Altman argument that training AI is comparable to training humans.

It also takes a lot of energy to train a human. It takes like 20 years of life and all of the food you eat during that time.

This is either intentionally dishonest or genuinely confused. Let us do the math:

• Training GPT-4: Estimated 50,000 MWh of electricity
• Average human lifetime energy intake: About 2,000 MWh

The comparison makes no sense. Humans learn efficiently through biological processes that evolved over millions of years. AI models require enormous computational resources for tasks humans do effortlessly.

The Real Problem

Here is what Altman gets right: we should absolutely be concerned about AI total energy consumption in total, because the world is now using so much AI.

He is right about that. And his solution—move towards nuclear or wind and solar very quickly—is actually reasonable.

But there is a contradiction here. If Altman truly believes we need more clean energy, why is he dismissing concerns about AI current environmental impact?

The answer is simple: because acknowledgment creates liability. If AI companies admit their environmental impact, they become targets for regulation, backlash, and bad press.

What This Tells Us

When a CEO best defense is but humans also use energy, you know the industry is on shaky ground.

The AI energy crisis is real. Water usage is real. Rising electricity prices in communities near data centers are real.

Altman deflection does not make these problems go away. It just makes him look like he is more interested in protecting OpenAI reputation than addressing real environmental concerns.