Ep9. GPT-4o, Astra, Multi Modal, China Tariffs, Tech Earnings | BG2 with Bill Gurley & Brad Gerstner

This is classic freshman economics. If someone's better at doing something than us, we should let them do it. We should buy it from them, and we should send them what we're good at. This is just deglobalization, highly inflationary, and will make our companies less competitive, not more.

Hey, man, great to see you. That was fun to take some of your money in Vegas this past weekend.

I can't believe you're bringing that up.

Last night, I'm sitting on the back porch. I'm helping Lincoln study for his macro-econ test today. Hopefully, he did pretty well on it. I have to say, as a dad, this is one of my favorite things, just watching kids gain that agency, really get hungry and curious. He was asking questions about the IMF and the World Bank. I didn't know the precise answer to them, but I just opened up ChatGPT, and I just found myself—ChatGPT literally became, you know, almost like a teacher who was sitting at the table, non-intrusive, uh, answering questions for us, keeping the conversation flowing, you know, and I found myself, we never opened Google a single time, Bill. Um, and it really unlocked a lot of learning and led to this 45-minute conversation back and forth that we had.

Um, and you know, uh, doing that and then watching some of the things this week, it really reminded me that unlocking education, both for kids like my kid, but also kids around the world who all basically have equal access to this incredible tutor on these mobile devices—I think it's hard to imagine the amount of human potential and kind of global economic impact that will have over time, knowing we're in the very early innings of this. Yeah.

I think it's a really good point. We talk about what can AI do and what can it not do. And I think there's a lot of fair conversations on both sides. But one thing that seems glaringly obvious is this teacher element. And Sal Khan, who's present in some of the demos this week, has been leaning into this heavily. He's someone we should definitely pay attention to. But they look real. And we talked about the models are infinitely patient, which humans struggle with. And I even tweeted when I was watching one of the demos, it reminded me of—there's a great Neal Stephenson book called The Diamond Age, where—and the subtitle of the book is A Young Lady's Illustrated Primer—and the narrative, the hero, heroine of this book,

gets a little tablet, and the tablet teaches her throughout the course of the book. And it's some really elegant writing, and it's very prophetic for what we saw this week.

Well, let's talk a little bit about what we saw this week because it was all about consumer. And, you know, in our first eight episodes, Bill, you must have brought up the movie Her, you know, five or ten times in its relationship to AI. So you must have felt vindicated this week when Sam Altman, you know, after the demos began, tweeted... her not to be upstage. Then Google yesterday, you know, at their IO event showed off Astra, their version of her, albeit a little recorded and maybe a little bit more clunky. So I made a little mashup for you. Yeah, maybe just as a bit of a bit of a demo to get us started here. So maybe Benny will will fire it up for us.

Good morning, Theodore. Good morning. You have a meeting in five minutes. You want to try getting out of bed? Get up! You're too funny. Theodore, I saw in your emails that you'd gone through a breakup recently.

You're kind of nosy.

Am I? You'll get used to it.

I feel really close to her. Like, when I talk to her, I feel like she's with me.

I want to learn everything about everything. I want to discover myself.

Keep going. Keep going.

Hey, ChatGPT, how are you doing?

I'm doing fantastic, thanks for asking. How about you?

Pretty good. What's up? So my friend Barrett here, he's been having trouble sleeping lately, and I want you to tell him a bedtime story about robots and love.

Oh, a bedtime story about robots and love? I got you covered. Gather round, Barrett. Once upon a time, in a world not too different from ours, there was a robot named Byte. What does that part of the code do?

This code defines encryption and decryption functions. It seems to use AES-CBC encryption to encode and decode data based on a key and an initialization vector, IV.

Happy birthday. So it only took us, you know, 11 years from Her to have kind of our Her moment. And I actually asked ChatGPT how long it usually takes from science fiction to it showing up kind of in its first iteration. And it turns out they estimate, you know, 15 to 40 years. So that seems pretty good, you know, in terms of the timeline there. But let's talk about what GPT-4o and Google Astra gave us this week. I mean, I think it's mostly these consumer-facing small assistants. The models are smaller, faster, but most importantly, it's like this end-to-end audio model, right? It can listen, it can think, it can speak in one model. The voice seemed to have a lot more empathy. It could interrupt.

It was reacting a lot faster. Still didn't have memory. Still didn't have actions. But the tone and the cadence of interruption, it all felt a lot more real. Yeah. What were your reactions?

So I agree. I found myself... feeling like both OpenAI and Google were tilting more towards the consumer. There's been a lot of talk. Is this about the API and enterprises? Is this about the consumer? Where are they going to make money? Where are they going to focus? These felt like, you know, I think OpenAI said something about the human-computer connectivity, which sounds like something Bill Gates would have said, you know, and very consumer-focused. I thought the winner, if you will, of the week was OpenAI's voice recognition. I will tell you that playing with it on my own phone, I didn't get the same reaction time you mentioned. I didn't get the same quickness; when I tried to interrupt, it hung.

So I don't know if they had an especially fast version or whatever, but, but I didn't have the same experience. I hope, I hope I will. Um, I think it's really killer, as I've talked about, when you think about voice, and if that's, you know, where they're winning, you got to think about what apps are voice-in, voice-out, you know, and, and, um, not all apps are that way. If you want to, you know, see four different hotel choices in Taiwan, I don't think you want that read to you. Um, I think that will be tedious versus looking at it on the screen. Now the combinations are possible. So there's all kinds of things we can do. Um, I, I, I wondered whose voice they trained these things on because the, uh, overt—

Excessive inflection was a bit cringy, as Elon Musk noted. I don't think real people talk that way. I was hoping maybe they'll put a smarmy slider in so I can take that down a bit when I use it.

Well, I think they actually demoed that at OpenAI, where you could turn up and down the dial, I think, on inflection, or imagine it's going to have the sexy dial, it's going to have the serious dial, it's going to have the teacher voice. You're going to be able to get whatever you want. But let's click on this voice thing, Bill, because I think it's really interesting. You and I have talked a bunch of times about kind of voice as the new graphical interface, right? And is this really the moment where, now that we have a voice, a model trained on voice that really is multimodal and brings these things together, you know, that we get to a level of latency and to get to a level of interaction that feels human enough

that it really does start to replace typing and the mouse. And I'll stipulate to you that it may in fact be that it reads me something and says, 'Hey, check out your phone for the four options of the hotel,' right? Where there's an interaction that's more blended. There are definitely going to be things that we're going to want to see, you know, but think of this, you know, design for the human world, but also AI, you know, enabled. So we're not going to throw out all the apps that we currently have like DoorDash and Booking.com, but where your AI will be smart enough to navigate those and then maybe take screenshots and push them to you or things like that.

Well, look, or put them on your—I think it's a really important point in time. And if they achieved what you said, I think it's like the first door that opens for us to go figure all those things out. It might be useful for someone to, you know, we got all these scores to measure AI. It might be useful to someone to come out with a score for AI voice recognition. I mean, there's all kind of different dialects people have and different types of voices. It'd be killer if there was a way to kind of gauge. It feels like from the demo, this is way better than Siri. Like I said, my experience wasn't quite up to what we saw on the TV, but it might be useful to know that. And if it is achieved, if there is this

this voice recognition that doesn't miss a beat, like in the movie Her, then I think that's a really important point of demarcation in time that that's been achieved. And I think it will allow all types of experimentation into the future.

Let's touch on that. You mentioned the benchmark. So in terms of the benchmarks, this wasn't actually a huge leap forward. You know, maybe that's why they didn't call it ChatGPT-5. Maybe because on certain benchmarks, you know, Llama 400B is already better. But it was accelerating at other things. So we have a couple of charts here. One is... We just took the SemiAnalysis chart, which is kind of this quality versus inference API pricing, and we plotted GPT-4 Omni on this chart, and you can see how it barely improved in terms of HumanEval score, but it dramatically improved in terms of pricing. You know, in terms of inference pricing. And then if you go to the next chart, this is just Elo performance versus release date.

Again, you know, ChatGPT-4o is an improvement. But it does appear on those measures, Bill, we have some maybe plateauing going on. But I think the point you made is the more important one. And this is one Karpathy tweeted about, which I think is interesting. He said, this is the first time that we really have a model that reasons across voice, text, and vision in a single model. It's processing all three modalities in a single neural net. Right? And if I just think about how humans operate, we're processing vision, we're processing audio, we're processing text and voice all in the same neural net. And so it feels to me like the path to reasoning, the path to feeling more human-like, much like the GPT moment we had with FSD V12 and self-driving, that you kind of had to throw away how we got here.

And you have to focus more on, you know, these, you know, bringing these modalities together. So it may be that the benchmarks, Bill, plateauing are consistent with, you know, having to tack to other ways to get model improvements, to get performance enhancements.

You may be right. And just to clarify, I think you know this, but just to make sure the audience does, I was implying maybe we need a new benchmark to measure voice interpretation quality. So that would be different than these, which is obviously an area where they improve. So you could be right. I mean, the notion of multimodal or mixture of experts or any of these things that have multiple models, might unlock different areas of improvement. It does strike me that the models and the scores kind of send everyone running in the same direction. And, you know, we've got in both of these announcements, you've got CEOs or someone putting up on a screen, you know, context window, you know, and two million tokens, like the general public has any idea what that means.

So it's useful if things move in a different direction, I think, and things splinter. One big takeaway I had from this is if they're all going at consumer, and then we had this announcement that Mike Krieger from Instagram fame is joining Anthropic. If they're going, and then some people rumored that some of the executive changes at Amazon are to get them more engaged on AI, we're just going to have six companies all running at the same prize. And, and it's interesting, we end up talking about AI every week, but I don't know what there is else to talk about because that's all anyone's talking about. And so we're gearing up for a battle royale. That's what it feels like.

Let's talk about this on a couple of dimensions, because first, if it's all about voice and all about assistance and all about consumer, then it seems like it has to be all about Apple, right? I mean, that's what we're all carrying around in our pockets. It brings us back to the rumors that they're going to announce a deal with OpenAI at their Worldwide Developers Conference in June. But as I've said many times, I think a decision to sell the generative AI search button to OpenAI and the way that they're doing with Google search is a far cry from outsourcing their mission to build their own version of Her, which I don't think that Apple can ever outsource. I think it's too existential to them.

But that brings us back to kind of like what will really make that special. And it feels to me like what was left out this week, right? What Her lacked was memory and actions, right? This ideal that you remove the final wave of latency on these devices by putting it on the device, a smaller model that can remember things and have actions. Now, I think the question that Apple's probably struggling with, Bill, is in order to do that, even a 7B model, you probably have to have 75 gigs of DRAM on your phone, the devices, and it's got to run everything else on the phone. And so I think that we're going to see a lot more memory that gets pushed to the edge. You're going to see these models continue to get smaller.

But a question that I have is, is can you package the multimodality that we just saw with GPT-4o? Can you package that into a small enough model, a Llama 3 7B, that we can actually sit resident on these devices?

Yeah, I'm sure that's a challenge. One that will be solved in time. That's not something that's impossible as time moves forward, but it might be impossible now. It's going to be super interesting to watch them try. One of my big takeaways from Google I/O was, and we had talked about this a few episodes back, that Google just has a lot of assets, you know? This is one of the problems with Anthropic and OpenAI is they don't have a lot of assets, but Google has a ton of assets, and it really showed up in their demos. They were touching YouTube. They were touching Gmail. They were touching Docs and Sheets. It was interesting to see Docs and Sheets come back to the forefront, which there is a world, and I won't suggest that this will happen, but there's a world where they get AI so right

that people move to Gmail, move to Docs, move to Sheets, move to the Google stack, move to Android, because the integration is so amazing. They're really the only one that has all of that.

There's no doubt about it. And I was watching Liz Reid. She was running the search demo for the new Gemini-powered search. I think she runs all search now at Google. It was doing exactly what we expected it would do, predictively doing this AI-embedded search, didn't get rid of the 10 blue links. You know, if you look at the sell-side notes out of banks this morning, Morgan Stanley and others, they're defending that this will be good for Google, that they've figured out how to monetize AI search. And so, you know, clearly Google has woken up. They've begun this path towards structural change as they talk about it and getting fit. You know, as I've said, I've gotten, you know, one toe on the bandwagon there because I do think that they have all these assets that you're talking about.

But I have to say the company with no assets, as you describe it, in the case of OpenAI, I would say they have more than no, but their assets are their people and their assets are they're small and they're moving fast and they got the funding from Microsoft to do it. And they obviously have Jensen delivering them the latest chips. So they got all the raw ingredients, but they're really pushing Google. And I thought that Google's version of its Assistant that they showed off this week definitely lagged OpenAI. But I also stipulate, I think there are going to be multiple companies like you said. I think Meta is going to get there. I think Apple is going to get there. I think OpenAI is going to get there.

I think that Google will get there.

Yeah, and obviously you know this, but by assets, I meant users, apps, data. I didn't mean the potential of the individuals. And there was one thing in the OpenAI demo that I thought was super interesting that, to me, will create tension going forward. They showed off a... desktop version of the app, which I don't think is out yet, but hopefully it'll be soon. Mac app. Oddly, a Mac app since Microsoft put $10 billion in them, but a Mac app. Might be some interesting reasons why that's true. But it looked at the other screens you had resident, and I don't know how it did that. I don't know if it's... just taking a screenshot and it has to be at the forefront or if it literally could look at any active app you have.

I do not know. A lot of questions on that front about how you do that and what you're allowed to do. It's interesting, like Google was built and Meta was built, you know, in, in a, in a Microsoft-dominated world, you know, then Apple came up, but they were able to insert themselves in the browser and then create enough power to rise up and do other things and create moats, Google with the Chrome browser, that kind of thing.

Right.

Um, it, it'll be interesting to watch OpenAI try and do that, um, because people are aware of what the possibilities are if you get to that place, if you get to enough breakout power. Google got to breakout power, launched a mobile OS, launched a browser, like fought back. So letting someone embed as a bit of a parasite is a risk.

Yeah, and I think what you're saying is if Apple were to allow ChatGPT to embed in any way within the iPhone, it's a little bit like AOL, who was the king at the time, allowing Google to embed and pull away a lot of search activity and kind of legitimize them. It's then very hard for AOL to kind of pull back search when they build it for themselves. Is that—

Yeah, or the examples I mentioned. I mean, you remember when iOS took off and really started succeeding and then Meta came public, Meta's stock went from 40 to 18. IPO'd at 40, went to 18. I bought it then. And everyone was worried that Apple just controlled too much of iOS and they'd be able to undermine what Meta wanted to do. I mean, that's another example. Now, it didn't happen. Meta got their stuff together. Mark says it was because he went public. It woke him up. And they got their mobile game together. And then they had to fight that war again with the ad network changes, and they've worked their way around it, but there is tension between a very ambitious app

on a platform. You know, well, I think, you know, and that's why I think when you look at kind of the business model you need to have to survive, right, because the inputs are so costly, it seems to me that everybody's going to fight for this consumer landscape. And the only—right, we, we have Perplexity out there with, I think, last we saw on Twitter, something like 20 million in revenues, and the people who love them really love Perplexity. But the fact of the matter is ChatGPT is the Google verb of the moment of AI. It does seem to me that they need to and are focusing more on consumer. I feel like it's a better path to AGI. And maybe just look at this cost per query of OpenAI's flagship model. You talk about these pricing cliffs in the enterprise, but the price is plummeting in the enterprise.

In fact, I was talking to a large data company this week who's in the business of serving models, and they were lamenting the fact that the pricing umbrella was set by OpenAI and there is no price, right? So that it's a much lower margin business than the traditional software business. So in enterprise, it's really kind of hard seeing anybody other than the big data platforms, right, kind of being in that game. So if you're going to be Anthropic or OpenAI or Mistral or anybody else, it's hard to see how you build a business with payback on the tens of billions that you're going to be required to invest unless you're in the consumer game. And the only brand right now that's really broken out in the consumer game is ChatGPT, other than the brands that already have big consumer distribution.

Yes. So a couple of reactions to what you just said. One, I'm fascinated by what I call the price cliff, which is just if you back off 0.5 of one of these release models, you save 90% or 95% of the money. It's like 20x differential. And when I talk to our entrepreneurs that are using these models, they might design with the cutting-edge model, but they all back off to the affordable models on implementation, inference, and runtime. They all do. There's no one that hangs around up there. And so it is interesting that that dynamic happens. It does raise questions about just how price competitive running AI models as a service is. I would love one day for you and I to have an episode dedicated to whether

I don't know whether cloud hosting is a good business or not. I'd like to find out. I'd like to talk to some experts. But it may be to your point. It may be that the enterprise side is just too competitive, too tough. We'll see. I think OpenAI believes that if they get voice right, if they get data right, memory right, that people will want to develop apps with all of those in there, and they will get more locked into their platform. That's TBD. We'll have to see what happens.

Well, there's no doubt that if you build, you know, and I think that's the bet they're making. And frankly, you know, listen, this is like, the reason we talk about this every week is only three or four weeks ago, Bill, that Lama was out with its new releases and it broke the internet talking about, you know, the breakthroughs with Lama 7B and the integration across the family of apps at Meta, et cetera. And it seems like this is punch, counterpunch, but... Unlike prior eras, it's not punch counter punch with just small, you know, underfunded startups. This is the largest companies on the planet who are all in, you know, in this game. And, you know, one of the things that I saw a lot of people talking about this week when Omni and Gemini upgrades came out was who does it kill?

Right. Like we're talking about who wins. But there's—there's been record funding into AI by venture capital over the course of the last two years. We've talked on this pod and others about how Inflection, who was building—right, remember Pi, which stood for personal intelligence—was trying to build a consumer app, built one of the largest H100 clusters in the world, raised a lot of money, then Mustafa said, "This is going to be too hard," sells the business to Microsoft. So Microsoft—he's now running Consumer AI at Microsoft. By the way, I'm not really sure yet what Consumer AI at Microsoft is, because I'm not sure what the brand is. You know, is it going to be Inflection? Is it going to be Bing?

Is it going to be this Copilot? But they need, you know, they need to sort that out. I'm sure Mustafa, you know, is going to work hard on that. But, you know, so Inflection's out of the game. Character, which we heard a lot about, raised a lot of money. You know, I was watching OpenAI's demo this week, and I said to myself, "That's kind of what I thought Character was going to be." They raised a lot of money. So where is Character in all of this? And then Sam said in this Stanford interview a couple of weeks ago, Bill, that if you're a GPT wrapper, then you better go find a new line of work because they're going to continue expanding the concentric rings, and all the wrappers, even if they continue to survive, the value they'll be able to extract out of the ecosystem will get smaller and smaller.

Yeah. So as you think about what you saw this week, I saw somebody tweet, I'll leave it here. He said, "Startups dead: sentiment analysis, live meeting assistant, translation apps, language learning apps, music/sound generation, tutoring apps, et cetera." Is that what this all comes down to? That it's going to be winner-take-most and they're going to eat a lot of that app ecosystem as well?

It's a big question. We talked about education. Chegg, which is a company that was, or is and still exists, heavily focused on helping college students make their way through their homework, it got hit early when these models started coming. People do ask the question, what's next? A lot of the demos at OpenAI were language-related. They had two of them that had—and, you know, one of them was a live translation where you just put the phone out there. I'll speak my language, you speak yours. And it recognized and interpreted the voice, recognized the language, knew the language of the other person, and just was a translator. And I took the phone out that night, and thinking about it and thinking about language tutoring and companies like Duolingo, I just said, "Hey, I'm a beginner Spanish student."

Could we do some lessons? And it just jumped right in. Just jumped right in. Like, sure. And we did the first lesson and it said, is that easy a level or too hard? I said, let's go up a level. And it just went up a level. And I don't... I don't think anyone coded the language tutor into this thing, right? It was, that was all emergent. So it's pretty powerful.

I mean, you have to, I think the day that the demo showed Duolingo stock was down, you know, four or 5%. You know, so again, I think, you know, one of the things you see, and we'll get to this later when we talk about kind of markets and multiples and valuations, but valuations are a discounted, you know, cashflow. It's the expectation about future cash flows, right? And so the one thing that happens when you have this level of disruption at this scale for so many businesses, the multiple you apply to those future cash flows just needs to be lower. Your margin of safety needs to be higher. So stocks come down not because people are like, oh, my God, this is going to take it out tomorrow. But the fact of the matter is Duolingo is valued on its next 10 years of cash flows.

And people are simply saying, I have a little less visibility. This makes me a little bit scarier, more scared that those subscription revenues that are going to Duolingo aren't going to be there. Or at least the level of confidence I had yesterday, right, is lower today based on these things that I'm seeing.

And that reminds me of two things. So one, I've talked in the past about my good friend Michael Mauboussin. He wrote a paper about 25 or 30 years ago about what he calls CAP, competitive advantage period. And that competitive advantage period is how many years of cash flows into the future are embedded in this company's stock price. And it's a measure of how durable Wall Street as a whole believes your business model is. It's almost a moat measurement. And so you're absolutely right. If, if there's a question about your strategic value, your multiple can come in big time. And that reminds me of the second thing I was going to say, which is a famous quote from a poker game long ago. Multiple compression is a bitch.

It is not fun. No doubt about it. To say it another way and to give an example, people often ask me, they're like, Coca-Cola may have low single-digit growth and trade at a higher multiple than a tech company that's also profitable and has a high rate of—and they're like, how could that possibly be? You know, doesn't the market, isn't the market mispricing those? I'm like, no, because, you know, back to CAP, market is saying that Coca-Cola is going to have those free cash flows, you know, forever. And what they're saying about this tech company is it may have them forever or it may not. I'm just applying a much higher discount to that probability. And you and I have lived through this before.

I remember vertical search engines when Google started pushing deeper into the verticals, right? All these companies that people love like Tripadvisor that absolutely got eviscerated because one day they showed up and Google was showing basically the content that previously like they had dominated. And so as a firm, as an investment firm, as you know that on the public side, you know, we look at who's going to be disrupted. I think, you know, this is the list of companies that we have that are potentially disrupted is as long as it's ever been. You know, we're coming out of this period of stasis, you know, where things didn't really change that much in consumer land. Right. And so, you know, there were—not since mobile, not since mobile.

Yeah. And that was, you know, 10 years ago. One little piece of advice I'd give people that are thinking about this, you know, I've said from the beginning, like LLM stands for large language model. So things that involve language are what this thing was designed for and what it's really great at. Really great. And also, before that, code is a subset of language that's actually more structured, and therefore it's even better at that. I thought my very favorite demo from both of these things was the one you showed in the video where someone just has their camera on a piece of code and asks for analysis of what's going on there. That is powerful. Like, that is useful. That is high-utility stuff.

And will help people be more productive, for sure.

Well, one of the areas, you know, so I think what you were saying is if you're in the business of language like Duolingo is, then you kind of go, you move from the confident category to the less confident category because people start wondering about this disruption. I'll tell you a whole 'nother category of companies, Bill, um, that have moved into that bucket, and it's all these BPO companies, business process outsourcing businesses. Right. So, you know, because remember, it's not just about language now. Right? BPO companies woke up yesterday and they're like, 'Oh my God, this is about voice. This is about call centers. This is about video in. This is about, you know...' And so all of those modalities now call into question the things that BPO companies do very well.

And so that's the type of disruption that I think is going to be rolling thunder. You can't just look at where we are today. You have to look at the rate of innovation. The rate of innovation is the only thing that matters here. And I've never seen rate of innovation at this pace. And so I think that the surface area of disruption, the companies that are going to get disrupted potentially by this, and the market will discount it before it actually happens. But there's a long, long list of those companies.

And I would just qualify one thing you just said. I think with language and code, this is already, you know, 10 out of 10, high alert. I think on business process automation, we think it's coming. I don't think it's at 10. Maybe the warning signal's at 6. It hasn't been proven it's going to run it over yet, and there's still some experimentation to happen. Yes.

One other thing I want to hit on this topic coming out of this week on AI before we move on is all about Washington, right? And there are two things I think that we have additional information on as it pertains to Washington. One is on the regulation side. The other one is on the investment side. On the regulation side, I heard a good discussion with our friends over at All-In and Sam on this idea we should regulate outputs instead of regulate inputs, right? The idea that don't, they shouldn't be in there mucking around with weights and everything else, but like a plane, you know, if you have a frontier-level model, you should have to show it to them. They get to run it and they get to decide, does this meet whatever safety standards?

When I was listening to that, and they seemed to kind of galvanize around this 'regulate outputs' regime, I was thinking about you. Does that really cross the line for you? Like, is 'regulate outputs' a good enough standard for you? And then how do you really implement that?

Yeah, I mean, it kind of depends, right, on how it's implemented. I think, you know, David Sacks had said there's laws already. So if you commit bank fraud with an AI model, you've committed bank fraud. You don't need an AI bank fraud law. The tool doesn't matter if you use a gun or an AI model, right? And so I do think that's a... the type of attitude that can cause everyone to take a deep breath, because I think the odds that any government in any country could get in this early and start messing around with inputs and be effective is near zero. And, you know, I think Yann also said, you know, these models have been out here for a couple of years. Like, where are all the crimes? And it's another just take a deep breath moment.

It turns out early this morning there was a large 20-page document dropped by Schumer in the Senate, which I guess some people were waiting on. And I read through it. It wasn't that long. And if you're up for it, I'll tell you what's in there. No, I think that's it.

That to me is really interesting because that's less about regulation. And I think more about government now deciding they want to invest a bunch of money in AI. So why don't you give us the breakdown?

Yeah, I mean, that was my big headline surprise. So, well, actually, I'll just read you. It was broken into pieces: Supporting U.S. Innovation, AI in the Workforce, High-Impact Uses of AI, Elections and Democracy—which was only two paragraphs—Privacy and Liability, Transparency, Explainability, and Intellectual Property, and Safeguarding Against AI Risk and National Security. So those were the things. It was very high level. It didn't recommend any kind of legislation at this point. The big shocker for me was it recommended spending $32 billion a year. And I didn't even know that was on the table. I thought we were trying to manage risk. I didn't know that the money needed to be handed out.

Certainly in my lifetime, and I think prior to my lifetime, there's never been a venture capital category that's attracted money. more money. So the irony that we would even need even more money is laughable to me. Now, it turns out if you dig into it, some of the money's more DARPA-like, like they want it to drip into the DOD, they want it to drip into the military, they want it to drip into academia. But even with that said, let me put this in perspective. The NIH is 40 billion, a little over 40 billion. They'd like it to be 50. The National Science Foundation is nine down from 10. So 49, you're going to start the AI foundation or federation or whatever it is at 49. Institute at $32 billion a year.

You're going to put it right below the Institutes of Health and 4x the National Science Foundation.

It just seems a little overkill. You said that they were wanting to drip money into these things. Let me just remind you, Washington never drips anything. It is a flood. It is a fire hose. And I just think, at a point where we're $38 trillion in debt, where now we're moving toward a trillion dollars in interest payments—again, when we think about trade-offs, there are certain places we actually need money and there are other places that we don't. But it's like right now, it certainly feels like we're presented with a long menu of things that you could potentially order at the restaurant, and they're just choosing all of them.

Yeah, and take the military. I've never seen more venture capital excitement about the military ever. So I just wish they'd take a deep breath and gauge what's going on in the private markets. We're awash in innovation and speculation and funding. And yeah, I don't think that's needed. Let me hit on three other things that are in there that are kind of interesting. They referred to some AI systems as, quote, "black boxes" and said this may raise questions about whether companies with such systems are appropriately abiding by the laws. This is a bit of a turnabout, because it's my belief, based on everything I've read and seen, that the proprietary model companies were in there begging for regulation and urging open source to be cut off at the knees.

This looks like it rebounded in their faces. And the academicians I've talked to have made this point, that the open-source ones are visible and transparent and these others aren't. So that was kind of interesting to see that in there. The second thing, they talked about using AI to improve efficiency in government. The reason I chuckle is, I think AI could be amazing at improving efficiency of government. I put a very low probability on them actually leaning into that Milei style, perhaps. 15% of the US workers, 45 million, are government workers. There is an unbelievable amount of opportunity. But I just, I really doubt they'll lean into it. And then, and then the last thing. And I think this is a huge positive, although I don't want to overstate it.

There was a quote that related to improving immigration for high-skilled STEM workers. Now, all it said was the relevant committees to consider legislation to improve. But it's nice to see that in there. That is something I think everyone in Silicon Valley has been rooting for.

No doubt. Hear, hear. And I will say I was out there a couple of weeks ago at an event, you know, great event called the Hill & Valley event. And there's never been more engagement between Silicon Valley and Washington, D.C. And I think it's proactive engagement on the issues that matter most. I hope it doesn't result in a flood of money because I think that the private markets are better at allocating those dollars, and a new shift into industrial policy by the United States where government is picking winners and losers is a bad idea. I don't think that's where we're headed. I think Schumer and his colleagues on both sides of the aisle are thinking smart about this. Jay Obernolte, who's leading the task force in the House, thinking smart about this.

So I left very optimistic about how I'm feeling about Washington getting how they're seeing it. I don't think we're going to get a lot of regulation. I would be surprised if we see a lot of incremental dollars go into this. And, you know, from your mouth to God's ears on immigration and government efficiency. But while we're talking about Washington, I saw another tweet out of you this week that caught my eye. So I want to talk about it because there are some other news. And, you know, Biden said, just a couple of days ago tweeted, 'I just imposed a series of tariffs on goods made in China: 25% on steel and aluminum, 50% on semiconductors, 100% on EVs, 50% on solar panels.' And, you know, quickly Twitter blew up because people posted his tweet from the then-Senator Biden when Trump issued, you know, tariffs on China where Biden apparently tweeted,

"Trump doesn't get the basics. He thinks the tariffs are being paid by China. Any freshman econ student like my son could tell you that the American people are paying for the tariffs. The cashiers at Target see what is going on. They know more about economics than Trump." So has something changed here, Bill? Do we have valid national security reasons today that we didn't have five or six years ago when it comes to imposing these obvious economic costs on American consumers?

Well, look, I mean, I totally agree with Senator Joe Biden, who apparently is a different human than President Biden. And I don't know, I really don't know how the media doesn't contrast those two things and make a big deal about it. You know, my tweet was just highlighting that he said—well, he said two things. He said, "I'm determined to ensure America leads the world in these categories." Giving someone a head start doesn't make them faster at the 100-yard dash. It makes them slower. It makes them less competitive. Someone replied to my tweet and said that Germany was—over in China—and used the word "fitness", that it makes them more fit to have to be engaged in the competition on the field.

And I mean, not only is there that element, like you want to be competitive, right? But the other element is, you know, this is classic, you know, freshman economics. If someone's better at doing something than us, we should let them do it. We should buy it from them and we should send them what we're good at. Like, this is just deglobalization, highly inflationary, and will make our companies less competitive, not more. One thing I did when I saw this that I would encourage everyone who's interested in this topic to do, just go on YouTube and look at videos that were made at the recent China Auto Show. And these companies are creating more competitive products, not just on AV, not just on landed price point, but even on features consumers care about.

Much bigger screens, more interesting features you've never seen before. It's abundantly clear that innovation is most alive and most well in the China market's auto industry, and putting up these tariffs to help protect our companies won't make them stronger, it'll make them weaker by not exposing them to the reality on the field. It also didn't surprise me that, uh, that Biden was surrounded by a bunch of people in union shirts as he, as he gave this, as he announced this. By the way, one little—sorry to go on and on—one little thing that I hate about the Biden tweets, and I don't think he's doing them. I think someone else is. He uses the word "I" all the time. Like, what grade in school are you told, if you work with a team, you should say "we" instead of "I"?

Like, it's so easy. Like, this is just amateurish to use that pronoun there. He shouldn't be doing that.

One of the things that you and I talked about before we launched the pod—we're not going to spend a lot of time on politics, but the intersection of the political stuff with U.S. free trade policy, U.S. industrial policy, with AI regulation and innovation goes to the very heart of what makes us competitive economically. And we are focused on what makes us competitive economically. And when I look at this, what I worry about, Bill—for 20 years, I think a lot of people forget the late '70s where we had a lot of protectionism in the United States, but we also had double-digit interest rates and double-digit inflation. The U.S. economy was losing its way. The Japanese automakers were ascendant.

The European economies were stronger than the U.S. economy. And then we had basically 30 years of unabated move toward free trade around the world. And the U.S. led the way for global free trade, right? And of course, one of the consequences we know of free trade is that it leads to dislocation. There are people in the United States who get dislocated because it is cheaper to produce things. So we had NAFTA and we had these trade wars that we resolved over those years, but we always seem to resolve them in the favor of free trade. And I just wonder if we're entering this new era—a new era of the end of free trade, more deglobalization, the rise of U.S. industrial policy where we're picking winners.

What scares me is that seems to have supporters not only on the Democratic side of the aisle, but it seems to have a bunch of supporters on the Republican side of the aisle, including President Trump and including a lot of people in Silicon Valley who, on all other issues, seem to be free marketers. But when it comes to these two issues, particularly as it pertains to China, they're quick to jump on the bandwagon. They say, 'level the playing field with China,' but it doesn't feel like that to me. It feels like we're the ones who are kind of leading the way on deglobalization. Any big-picture thoughts on that? I mean, it feels to me like it will make us less productive.

I'm glad you brought that up. I mean, that is the primary reason that this stuff matters to me. And in the speech I did on regulatory capture, I mentioned these two Matt Ridley books, How Innovation Works and The Rational Optimist. And The Rational Optimist walks through history and shows that rises in prosperity and standard of living are always tied to free trade and the free exchange of ideas. And if we start untangling those and moving away from those—and other countries have done this in the past; China once led as a global nation, pulled up the walls, and fell precipitously. You know, and yeah, I think the whole vilification of China thing's misplaced personally. But I also, you know, when I think about—

—human prosperity, I think the world is small enough now where you have to think about that on a global basis. And it's unclear to me why a worker in Iowa needs $40 an hour if there's a worker in Mexico that's willing to bust his ass for $15. I don't necessarily understand that being a higher moral ground.

Well, it certainly seems to me—politically, I understand why it exists. And again, like, these disruptions are hard to deal with, and we're going to have a lot of disruptions that come from AI. You're going to have a lot of people looking for more protectionism, a lot of people looking for more end to free trade. I think what's made us great as a country is we've resisted those things for the last three decades. We need to resist them into the future. Back to what you said: the greatest path forward is to be the world leaders in innovation. And that requires fitness with competing against the best, starting at the same place on the starting line, not having unfair government-imposed advantages.

And by the way, one thing that people... I don't know. I think everybody that listens to this podcast probably is aware of this. But, you know, there have been quotes from, I think, Mike Moritz at Sequoia where he's like, 'I've been hanging out in China, and these people are amazing. They're hardworking. They're smart.' And I would say that's true. Like, the time I've spent over there, I've adored. The people are wonderful. They're certainly as smart and entrepreneurial as anyone I've met here. And so there's not, you know, I think hanging our hat on the fact that we're losing because their government's supporting it—man, maybe it's because they're really good. They're really, really good at what they do.

It just seems to me that we've won with China over the last two decades. Okay, I'm the first to say if they're engaged in nefarious tactics, we ought to meet them where they are. But I think the middle ground here—we may have swung too far. And the middle ground should be the continuation of free trade and finding ways to work together. That's how you prevent wars, and that's how you move humanity forward.

Let's, let's wrap up with a, with, with a public market check. What's been on your mind lately, Brad, since we talked last?

Well, you know, maybe we start just with the news out this morning because it's a segue or a bridge back to where we talked about before, which was what's happening with inflation and rates. So we had a CPI print out this morning, Bill. CNBC said, you know, the CPI print came in quite dovish, 20 basis points. Sequential cooling in the rate of change was the largest in several months. Retail sales came in pretty weak. Sales X auto and gas came in pretty weak. So you still have inflation-adjusted rents coming in pretty hot. So CPI gave us a little bit of a break. I think the 10-year is down to 435 this morning. In fact, you can look at this chart here that we have that shows the restrictiveness of

So where are interest rates compared to inflation? And we're really crossing this line right now. So the yellow line is the proxy Fed funds rate. So this is like what the San Francisco Fed calculates as the actual interest rate paid by Americans. The black line is the 10-year. So you can see that the proxy Fed funds rate went above inflation earlier this year. The actual interest rate is going to go above the rate of inflation. The forecast we have in here is simply the consensus Morgan Stanley and Goldman Sachs forecast for the balance of the year, which has been pretty accurate. So what is this, you know, telling me? The backdrop around inflation continues to be constructive. We had a couple months in there where I think people got scared.

Larry Summers said maybe the next rate move is up. Why does that matter so much, Bill? Because if people can earn, you know, just listen to the Berkshire Hathaway annual meeting. You know, Warren Buffett says, listen, I'm collecting my interest payments every week. If I can earn 5.4% taking no risk, why would I take risk? So I still am in the camp that we're on a glide path. It's going to take a little bit longer. But I do suspect that rates are going to come down. Jay Powell said yesterday we're on hold for a couple more months. I think you're going to get a rate cut before the election. But I don't think the market actually needs a rate cut, Bill. What they need to know is that inflation is coming down and the Fed can give us a rate cut if they want to, right?

That's the important point. And so I think that backdrop was important this morning. But I juxtapose that, Bill, with what we've seen in this earnings season. So we're almost through the earnings season. We knew coming into the earnings season that GDP had come in lighter than we had expected for the quarter. And now here's the update on earnings. So this first chart, Bill, this is the S&P earnings on a quarterly basis, the growth, if you back out the Mag 7. So we know, obviously, we got this AI tailwind. So in the quarter, if you back out Mag 7, we're coming in below expectations, right? You're actually seeing earnings shrink on a quarter-over-quarter.

Is this... That's quarter-over-quarter? Okay.

Yep. And then if you look at the next chart, which is our quarterly performance of software companies versus consensus estimates. So there's a chart that Jamin on my team makes. You can see in the quarter, a lot of the companies hit—98% of software companies hit their guidance. But remember, they had reduced their guidance in previous quarters. So maybe the bar was just easier to get over. I think... the prospective view is this next chart, which is guidance versus consensus estimates. So what are they saying about the future, Bill, versus what consensus thought they were going to say about the future? And 54% of software companies disappointed in the quarter. And that's the worst quarter I think we've had since 2022 in terms of their view as to the future.

And so I did a little stint on CNBC last week. I said, you know, we've turned down our exposures a little bit, taken a little bit of our net exposure, so added to some shorts, taken some of our longs down a little bit. And people said, well, why did you do that? And I said, well, you know, look at the backdrop here. The NASDAQ this morning hit an all-time high, right? And I always say to my team, all-time highs, that's the highest in a long, long time, right? And the backdrop is that the economy is slowing, earnings are coming in lighter than people expected, and AI is creating more disruption, creating more uncertainty than we expected. So this is a moment in time where I think that if you're an investor,

You know, owning some of the companies that you think you're really confident are going to continue to compound and benefit from AI, the Microsofts of the world, you know, the Nvidias of the world, seems to me like a relatively safe place. Those earnings multiples don't seem to me to be too onerous, right? Between 20 and 30 times earnings for, I think, 21 times for the Mag-6 anyway, right? But I think you do have to look at whether you own retail names or a lot of other names that are not exposed to those trends. So long as we're going to have interest rates above the rate of inflation, be in this restrictive territory, the trend is down in terms of economic growth, which the government, Jay Powell, needed to manufacture slowing growth in order to get inflation down.

But I think that's the backdrop. I don't think there's any fall off a cliff here. But I do think that we're in this situation, Bill, where the Fed is probably at some point in time going to need to give us some relief on rates in order to keep economic growth in this kind of 1.5%, 2% GDP range.

Yeah, I hear you. And look, there's two other things that would cause me to share your opinion about being conservative at this moment in time, especially related to those things. One, Washington does not appear to be concerned about inflation, which means they're spending both Ukraine, Israel, this new AI thing. They don't seem to be concerned about expanding the budget, which isn't helpful, obviously, on inflation. And then second, it's a chaotic time. I mean, with, with, with colleges moving to the summer, we're losing kind of the craziness that was happening on campuses. But we do have this election coming up. We have one candidate in court potentially going to jail. Like there could be quite a bit of social unrest in the next six months.

And I think that causes you to want to be cautious also.

Yeah, there's no doubt about it. And this is really just a question of units of risk that you want to have on the table at a given moment in time. So you have the NASDAQ at an all-time high. You have all those concerns that you lay out. And we were just in Vegas. And it's like we got a couple more cards turned over. And the fact of the matter is all the odds got a little worse for us. It's not to say that we're necessarily folding the hand and getting out of it altogether. But we're not pushing all in either. I mean, this is just a time, I think, to have a portion of your stack working. And if the market happens to go against you because we have a bad event happen at the DNC convention in Chicago or because inflation bumps back up because of all of this fiscal spending that you point out, then you're in a position to do something about it.

And so this is, you know, I always talk to our team just about less versus more. And we're in a moment in time where maybe a little bit less makes sense. It's great to see you. Great conversation as always. Thanks for making it happen. Take it easy. As a reminder to everybody, just our opinions, not investment advice.