Episode 6 — "The Thirst"
A number's been sitting next to Bear's whiskey all week: the claim that a single AI chatbot question drinks a whole bottle of water. He's not ready to tell you if it's true — but it's the thread that pulls open everything else he doesn't understand about the machine he talks to every day.
So he brings in Watts — Dr. Wattson Castor, a beaver, twenty years in hyperscale datacenter thermal and power engineering before moving into ML infrastructure. If anyone's built a career moving water and power around for a living, it's a beaver.
They cover:
- The 500ml-per-query myth — what's real, what's exaggerated, and the actual number (10.9 billion gallons, just Google, in a year)
- Why the water question really starts inside the chip fab, before the datacenter even exists
- Waste heat — can it be turned back into power? What Stockholm and Dublin are already doing with district heating
- The real grid number: US datacenter power demand jumped from 31 to 41 gigawatts in a single year
- Nuclear power's actual timeline for AI — real contracts, real restart dates, no hype
- Datacenters in orbit — is it sci-fi, or already happening? And the line that reframes the whole idea: space isn't cold, space is empty
- What a "token" actually is, and why agentic AI tool-calling can cost 4x the energy of a single answer
- Why the best coding AI in the world can't draw a picture — and why that's not a flaw
- AI and jobs — the ATM-teller parallel, and what's actually happened to entry-level hiring
- Quantum computing — where it really is, and whether it changes anything about how AI gets built
- A quieter turn at the end: what we built to answer our own questions, and the ecosystem none of us fully control
No lecture. No side. Just Bear, an engineer who can't stop fixing things mid-sentence, and a straight answer to the question everyone's been half-asking all year.
---
He Is A Bear is a fully AI-generated podcast.
Every frame. Every voice. Every word.
Built by one person at FroMade Studio.
New episodes weekly. Pull up a chair. Pour yourself something worth drinking. Let's talk.
----------------------------------------------------------------------
TAGS
----------------------------------------------------------------------
ai energy use, ai water use, datacenter power, ai infrastructure, tokens explained, quantum computing, ai and jobs, he is a bear, podcast, ai podcast