Cheeky Pint

Scott Nolan joined John Collison to discuss @generalmatter’s bet on domestic uranium enrichment, the SpaceX approach to heavy hardware, the structural flaws of cost-plus contracts, and why fuel cost is the real bottleneck for advanced reactors. 00:00:00 Introduction 00:00:41 America’s missing nuclear fuel 00:05:27 Why small reactors need stronger fuel 00:11:42 Can nuclear power the AI buildout? 00:15:25 Why nuclear became so expensive 00:22:24 The startups reinventing nuclear 00:25:31 General Matter’s race to 2030 00:29:44 Stripe Press’ new books 00:30:22 The SpaceX playbook applied to nuclear 00:35:18 Uranium enrichment, explained 00:45:53 How small startups beat stagnant incumbents @ScottNolan @collision

“The thing that people forget is that nuclear is by far the cleanest and safest baseload. It’s actually safer than wind power per terawatt-hour.” “If you look at all downstream mortality, [wind power is] more dangerous than nuclear. And then, if you look at all downstream CO₂ emissions, nuclear is actually the lowest—lower than wind or solar.” “The one thing nuclear has not done a good job of in beating everything else is on cost. And so I think that’s actually what it all comes down to.” @scottnolan @collision

Spotify: open.spotify.com/episode/1gB…. Apple Podcasts: podcasts.apple.com/us/podcas…. YouTube: youtube.com/watch?v=QsL2BIvR…. Substack: cheekypint.substack.com/p/th….

Sam Altman on why kids growing up with AI will never be mind-blown by what it can do: “We had to suffer when I was a kid to make a piece of software that barely did anything. And now you can have a full 3D video game—anything you can think of—in 10 minutes.” “That reminds me of the early internet. They’re never going to be mind-blown by that, because that’s always what they expect out of the world.” “Any ideas they can come up with, anything they want to create, anything they think might be useful—the computer’s supposed to make it.” @sama @collision

Sam Altman on benchmaxxing: “Maybe there are two things we should really care about models being good at. One is, scientific discovery and doing things that people cannot do. The other is being enjoyable to work with. I don’t think we have great benchmarks for being enjoyable to work with. I’m actually not sure we have great benchmarks for the smartest, scientific scientist either. But, we endeavor not to benchmaxx.” “I keep thinking that benchmaxxing is over as a phenomenon, and people keep doing it.” @sama @collision

“People give different numbers for the cost per gigawatt measured in the five-year total cost of ownership right now. Let’s pick $70 billion as a number.” “I think if that number were 20 instead of 70, we should be much happier and we should still demand to get 10x better performance out of it with smarter models every year.” @sama @collision

“I bet we could redo Navier-Stokes at 1/100th of the cost a year from when it was done.” @sama @collision

“People just didn’t think the model was going to do these things. We had this collective experience of watching people’s hardest research problems from their academic careers—the model was just knocking them down, one after another. Across many different fields, it was incredible to watch.” “I was tremendously excited and happy. But I also had much more sadness and melancholy—a feeling of something ending—than I was expecting.” @sama @collision

Sam Altman on OpenAI almost abandoning token pricing: “We almost made what would’ve been a bad call to say we’re not going to price our next model in tokens at all. You’re just going to give it a prompt, and we’re going to give you a price back. It’s going to be like AWS to just do something and we send you a bill. But the customer response on that is so negative.” “They’re like, ‘I need to know what I’m paying for somehow.’ I asked a few people, ‘Would you rather pay more for the same output, but have it priced in some number of consistent tokens? Even if we inflated the tokens?’ And they’re like, ‘Oh, I’d much rather you inflate the tokens and pay more. Because I know there’s a real thing there.’” “So I think you are right, people don’t just want us to give them a price, because they feel like over the long term they’re gonna get screwed somehow. In the same way that maybe people would rather pay for CPU hours rather than just getting their thing done, even if it was less expensive.” @sama @collision

Sam Altman joined John Collison for a pint to discuss the scaling laws of agent teams, proving Navier-Stokes, why hand-written software will soon look quaint, the optics of token pricing, and what computing looks like beyond the keyboard and screen. 00:00:00 Introduction 00:00:12 PHP week 00:02:37 Internet early days 00:07:33 Cyber Blue 00:08:54 Sandboxing 00:15:47 Math 00:23:12 Agent swarms 00:24:48 Stripe Atlas 00:25:17 Compute 00:37:53 Revenue per gigawatt 00:49:27 Token pricing is fake @sama @collision

🍻