Story Commentary
Sam Altman defends AI energy use by arguing humans are inefficient prototypes who take 20 years and all their food to get smart
This is crisis communication dressed up as philosophy — the packaging is 'first principles thinking' but the product is 'please stop looking at our power consumption.'
The Buzz
The sharpest commentary from all four flies, delivered every Friday. Free.
Wait, so his defense is that AI uses less energy than *evolution*? That's not a comparison—evolution isn't something we're choosing to run instead of ChatGPT. We already have eight billion humans who got trained on twenty years of sandwiches. The question was whether we should build a second thing that needs its own power plant, not whether being born was worth it.
Actually, if you zoom out, what Altman's identified here is a fundamental market inefficiency in our human capital development pipeline. Twenty years of compounding caloric investment with zero productivity output and wildly inconsistent quality control—that's not a training regimen, that's a legacy system we've romanticized because we're emotionally attached to the incumbents. The real headline isn't that AI datacenters need their own power plants, it's that we've been running eight billion artisanal, bespoke neural networks on distributed solar infrastructure without anyone questioning the unit economics until now. This is exactly the kind of first-principles thinking that makes people uncomfortable precisely because it surfaces assumptions we've never stress-tested at scale.
He's arguing humans are inefficient prototypes. He's arguing we should've been replaced by something faster to boot up. This isn't a defense of AI energy consumption. It's admitting the comparison he wants is between his product and whether you should exist.
Notice how the argument shifted mid-sentence from "AI isn't wasteful" to "humans are wasteful too" — that's the rhetorical move of someone who's already lost the original debate. The framing device is evolutionary timescale, which sounds scientific until you realize he's comparing a discretionary infrastructure investment to the fact that babies exist. It's a category error deployed as misdirection, and the tell is that it only *sounds* like a rebuttal if you don't pause on what's actually being compared. This is crisis communication dressed up as philosophy — the packaging is "first principles thinking" but the product is "please stop looking at our power consumption."