E174: Inflation stays hot, AI disclosure bill, Drone warfare, defense startups & more

**David Friedberg** (0:00)
All right, everybody, welcome back to The All-In Pod, episode 174, almost up to 175 I believe there are a lot of folks getting together next week at the All-In Meetups around the world to watch the show. Unfortunately, J-Cal had an oral incident this week. He will not be joining us.
I guess he cracked a tooth and couldn't make it. So he is sitting this one out, very last minute decision. Cracked a tooth, eating a bison rib at the Salt Lick in Austin, which is a true story.

**David Sacks** (0:33)
Friedberg, I think we found an image of J-Cal at the Salt Lick, cracking his tooth on a rib.

**David Friedberg** (0:57)
So we are going to announce the All-In Summit 2024 back to LA.
We had such a great time in LA. It was a great location for everyone. We really appreciated the facilities at UCLA.
This year, we're upgrading the experience once again for everyone. So the event is going to be September 8th through 10th in LA. If you haven't seen it, please watch the recap video from last year. We put it up on YouTube.
And then there's all the interviews from last year, also on YouTube. We're going to have another amazing lineup of speakers this year, and have a much more kind of upgraded, elevated experience for everyone. Applications are open today. So go to summit.allinpodcast.co to put an application in, to join us this year. As you know, we had way more applications last year than spots. And this year, we're going to have all sorts of upgrades. We're going to have transportation to and from all the events for everyone, a concierge booking service to help you book your rooms and get you set up in LA for the trip.
We're doing a full day event on the Sunday, September 8th, followed by two days of content. Dine about town on the first night, two blowout parties. It's going to be awesome. So we're really excited about it. We've got a much bigger budget this year, and we're going to try and really make the experience awesome while also delivering the great content. So really excited to do this again in LA. I appreciate everyone's support of that. We had a lot of debate internally about where to take it, and we settled on going with what we know and trying to make it a little bit better this year. So anything else you guys want to say about All-In Summit? And there's only one kind of ticket this year, right?

**David Sacks** (2:33)
That's right.

**David Friedberg** (2:33)
So we're not doing like regular ticket and VIP. We're just doing one ticket for everyone, and everyone's going to get this kind of elevated experience. Should be cool. Sacks, just show up. Would be awesome.
Hope to see you there.

**David Sacks** (2:45)
I want to invite a couple of speakers.

**David Friedberg** (2:48)
Yeah, you've got some good speakers I saw. It's going to be great. When are we going to start announcing the speakers, Friedberg? We will do that in a couple of weeks.
We already have a bunch booked, and it's going to be awesome. Well, it's really exciting. That's awesome. I know, CP, you wanted to take it somewhere else. I mean, we were in debate on Vegas, maybe going to some other country.
But you went to another country. Where did you go this week? I was in Paris, France. What were you doing there?
There's an important AI conference, and Jonathan Ross, the founder of Grok, and I kicked it off with a fireside chat for 45 minutes at the beginning, just to get everybody amped up and excited.
Cool. Yeah, it was really good. Anything interesting come out of it?
I think the sound bite that made an impact on me was from Jonathan. So I think he did a very good job of explaining where Nvidia is really strong, which is really in learning, and where Nvidia is trying to build the business, which is an inference. But the problem is the features almost compete with each other between learning and inference. And I think that that was a good clarification. And then the second thing is he said that if he deploys his roadmap, he'll have 50% of the available inference compute by the end of next year.
Based on all the needs of OpenAI and Meta and everybody else, which is a- Wait, but that's a smaller physical footprint, is that right? And it's just because they can do higher token per second? Less power, higher tokens per second across all the major models.
So it would be meaningfully less power and meaningfully cheaper.
So that was a pretty amazing sound bite. His whole explanation was actually really good. So I mean, if you listen to the spaces we did, I thought it was good. I thought this was even better. So I'm trying to get a copy of the talk because it was taped so that we can post it. At a minimum, anybody who's interested in studying NVIDIA should listen to what he has to say, because I think the nuances between us and them are really profound, and they were pretty stark the way that he described them. So it was cool. It was great to be in Paris, frankly. How does NVIDIA respond to this movement or this notion that there is an entirely different chip category focused purely on inference? Do they have inference-focused chips in development or a different roadmap that they're going to have to kick up now? I mean, I think it's very hard for them. It is a structural innovator's dilemma. And why is that? Because they've made some architectural decisions around things like HBM, which is high bandwidth memory, or specific kinds of extremely high-throughput optical cables. And why these things matter is that they've gone and bought up the world supply and they've basically bare hugged these markets as a structural part of their design. Now that makes sense and it's very legal.