Joe Wiesenthal on tariffs, the Moody's downgrade, rising global bond yields, and why the AGI narrative feels like a business story now

you so much for coming on. This is fantastic chat. We love to talk to you again soon. This is great. Have a great rest of your day. We'll talk to you soon. All right. And up next we have Joe Weisenthal, the stall war. 30% week over week. Pretty big. It's Wednesday. It's Wednesday. Uh well, welcome to the studio, Joe.

Bring him in. Bring him in. Bring in Joe Wis. I don't want to wait anymore. Oh, welcome to the show. Thank you. How you doing? Thank you for having me. I'm really good. I'm thrilled to be back here. How are you holding on? What are you doing on the market? The market was flat on Monday and we didn't know what to do.

Normally, we're either depressed and praying for something to change and then or celebrating a big update. We've seen these like 5% swings and it's just been boring. It's so painful. I thought like, you know, the market stopped moving and I was like, "The technology brothers will never have me back.

They'll never have a reason to call me again. " My I had I had I had one month of relevancy. People were listening to my podcast because the lines were moving up and down and then they stopped and I'm like, "Okay, well, I enjoyed it while it last. " Back to obscurity.

Yeah, we're we are having you on to We want to have you back on the show. We're having you on to move the market. Yeah. Yeah. This will drive the market. We want to create some volatility just off of some. Well, I mean, so so with the trade war, uh, we Yeah. Yeah.

I I I kind of had this like somewhat controversial take that like the the trade war was not like COVID or the or the interest rate crash because the the person who caused it could immediately undo it and Trump could roll it back which he did which he did and I was talking to Jordan Schneider trying to talk about this like he was saying like no this time is different it's not going to be like the trade war from Trump won we're going to look back on this as like the beginning of the end it's terrible he was very doompilled but it feels like now if we flash forward four years.

We could be looking back on this and being like that was a crazy but it wasn't that big a deal in like the world historical context. What's your read on it now? It's a really tough call because I was very sympathetic to that and as soon as he walks it back you're like you're just going to walk it back.

You're so stupid for everything this time. All that being said, we have a laugh track now. I love that. It's so professional. All that being said, a 10% baseline tariff on the rest of the world is not nothing. We are now a high tariff country. Oh yeah.

Um there there is this pause 50% tariff or whatever it going to be is not nothing. These not complete nothing and I do believe that they will change the economy in some ways that they anticipate. Look when he unveiled that chart on April 2nd I thought it's like well he's basically unplugging the economy.

We can't function with these levels. And he he sort of realized that too. He said the bond market got yips. He went back and then uh he probably heard stories about potential empty shelves of places like Walmart and Home Depot and stuff and then he dialed it back further. We're still in a new era though of trade.

It might be more like Brexit where it's like, yeah, if you go to I was in London a few weeks ago, it's still the same old UK. It's still pretty like rich country. It's doing well, but it's they haven't grown that much. It hasn't been particularly good.

So it might be the kind of thing where we don't notice it dramatically except maybe it's we don't grow as much as we could. Who knows though? Can you I don't know if it's the weather but when I'm in when I'm in the UK it's something about the slow growth of the economy that's just you can smell it. You can smell it.

Well well hopefully it doesn't cover America. Uh can you can you tie this to the credit downgrade? Uh how serious is that? Should we care about what Moody's thinks about? I don't I don't care about credit ratings. I mean, I'm of the strict view that we borrow in currency. We borrow in dollars and print dollars.

So, we'll never default. There is a you know, I would say there's two things which is one that um you know, it looks like we're going to get some sort of inflationary mix potentially between tariffs and tax cuts.

So I that's tends to not be very good for treasuries and to the extent you know that like policy changes in the US are now everyone understands that they can be unilaterally changed on a whim probably not great all things being all things considered for the perceived institutional stability.

So that's probably also not great uh at the margin for the international proclivity to invest in the United States. You know the one thing that's really notable is that a lot of markets clearly have rebounded to an extraordinary degree. Most notably the stock market. Yeah. Um the dollar is still very big. Bitcoin high.

Bitcoin alltime high. Amazing. Uh the dollar is still Well, can I by the way Oh, Pomp was on earlier. Can I just say a surprise guest? That's Can I just say um the the first time I was on this show, one of the questions the two of you asked me was, "Will Spax ever come back? " Yeah.

And I said, "Well, in my lifetime, they will definitely come back. " Pumped it has a new spec. When I said when when I said that spaxs would come back, I was thinking like 5 10 15 20 years. I did not think they said Pomp doesn't have the balls to launch a spa in the year 2025 and he proved them wrong.

He proved him wrong. I've never had a prediction come true that fast as that one. So, I appreciate Pomp vindicating uh my last appearance on TBPN. Yeah, it's great. Vindicated. Um, I want to talk about like like the Middle East deals, the what was your read on on all of that?

It didn't seem like it was super market moving per se, but it was very big in tech and it seems like there's this new eration there. It seemed like like amid all the the turmoil of the tariffs, there was like bright spot of like some really strong new alliances.

Obviously, there's some controversy there, but what was your read on the on the Middle East news? You know, I I I've been posting about this a lot, you know, since I hit uh my mid-40s, I've been reading a lot about the Cold War and mid- 20th century history because it's what you do at my age and so forth.

But one of the things that I've been reading about lately is that one of the, you know, every time the US and the Soviet Union appear to have some period of Dayton, so to speak, one of the entities or one of the forces that pushed the countries apart were the sort of like human rights groups, etc.

uh that really uh didn't like you know various activities uh going on in the Soviet Union and that often made it difficult to have peace. It's an interesting thing to think about well what if foreign policy just became a little bit more transactional.

We sort of accept that like there's like a very different system in this country than uh you know this other country. But you know what maybe we can have stable relations if we agree to sell them cuttingedge GPUs to set up an AI.

It's an interesting thought and I don't have like extreme or strong views on foreign policy. It's not my wheelhouse. Yeah. But it's an interesting thing to think about.

You know, for so long there was this view that you know like an important aspect of stable relations was some sort of common ground in governance and values and so forth. And maybe we are entering this era where stable relations are who we agree to sell chips to.

Uh on the other side, it's an interesting thing to think about. Jensen, another hundred billion for Jensen. Yeah, he's been on a terror. Jensen keeps winning. Yeah, except in China. Have you been tracking how many conferences he's been going to? He's he's it feels like he's at a different conference every single day.

And yet, normally that would be very very distracting for a CEO, but every time he goes to a conference, he leaves with a hundred billion dollar deal. So, it seems like it's working. It's really It's really remarkable.

And you know again I don't know how much like I've been trying to think about this like because the way people are talking about AGI and you guys have probably a lot of guests who know way more and being out there is like you know the race to achieve AGI is something akin to like which country will develop the nuclear bomb first we must get there.

I I don't know it feels a little sci-fi to me. I don't know how much to take that seriously. Other people are better equipped to talk about that. But there is this sort of like simple business storyt.

And so to the extent like just from a business standpoint, Jensen going around the world making all these sales potentially boxing out Huawei uh from the Middle East just from a business standpoint seems to be doing very well. Yeah, we were just talking to Scott Woo and uh from Cognition.

They have an AI coding agent and this idea that we have AGI, but it's just 15minute AGI. Like you can you can get an agent to go and do 15 minutes worth of research for you and compile a little uh a little dossier for you in deep research.

You don't have AGI to the extent that you can just say go create economic value for the rest of your life like you can with a human and the economic incentives just lead to a person finding their way in the economy and finding a job and buying things and consuming things.

I still haven't had like a great experience yet with any of the agent tools. I was um someone asked me they're like do you have a list of all of the Twitter handles of guests that have appeared on OddLotss? Um and I was like no but let's see if I could do that. So, I went to Manis, the Chinese Manis.

im, and I said, "Can you put this together? " Oh, yeah. Sorry. I know. Um, and then it's pronounced Manis, too, by the way. Manist. And then it it it worked for like like it was like 30 minutes later. I went back and checked and it said, "Oh, you're out of credits. " So, I did I never even got to see if it worked yet.

I don't know. I have yet to experience. Well, the question for me is is how long would do it take, you know, like a like an intern to do that job?

Well, it take a really long time, but like I don't even like but you know like my experience I think we may have talked about this last time like my experience like I'm still not totally convinced that they're like reliable yet. I don't know like I I I think you make a good point.

It's just that uh when I think of 15minute AGI I think that what's a 15-minute job? It's look at the most recent OddLot show notes and find me the one Twitter handle for the guest. Yeah. And now if you want to do that 200 times, that's not 15inut AGI. That's that's one week AGI.

And you need to spawn all these different processes and orchestrate them and and fact check them and build that up. And so yes, man should be able to do that, but it's not I think it's just wildly unreliable right now. And so sometimes you get an output that is 10 times faster and than your best employee.

And then sometimes it just completely fails. And we're not really equipped to like process. Okay, is the agent good? Okay, well sometimes the agent is incredible and then other times it's like yesterday I wanted it to count a bunch of boxes and it like worked on it for 14 minutes and failed. I completely agree.

Like I'm I'm every day still. And by the way, uh I've the music app I'm a songwriter. That's incredible. That actually of all the AI tools that I've played with, that might actually be the one that's most incredible.

I've been doing this thing where I just sing a song into my phone and then I say now produce this to sound like an 80s country song and it actually works. So that like some of these things just like absolutely blow me away.

I have a thesis for that because with w with music there's no right answer like there's no there's no oh you got one character wrong in the in the Twitter handle and it's not going to resolve right. But if you get one note slightly wrong but it still sounds like music.

you can kind of chalk that up to the vibe or or slight differences.

And so I feel like that's why the the image generation it's less like binary or you have code where it compiles and it can work against itself and check okay does the code work or not try again try again try again and so you either need to be when you're in the messy middle of like human management decision making analysis that's where AI is really struggling right now so just going back to the question of like can can AI compile all of the Twitter handles from every guest I'm guessing the answer Yes.

On the other hand, if that's what AGI is, I'm not really sure it's existential whether like China the Chinese model is better than open AI model. AGI is when you chat GPT to bring down the long end of the yield curve because it's really it's really getting up there.

And right now, if I could just go chat GPT and say, "Hey, let's do something. Can you help us out here? Let's do something that we need to that we need to develop right away. " Yeah. Yeah. So, let's let's talk about this. I I know you have a project level urgency. Yeah. Yeah.

Uh I know you have a a podcast episode coming out tomorrow about uh about bond yields globally, what's happening. Could you give people a little teaser preview on on on that and then they can go listen to the full thing? Absoluter standard chartered, one of our favorite thinkers. We actually never had him on the show.

you know, the story there's there's a there's a few moving parts to what's going on, but a big thing is this combination of countries around the world are spending more money. They kind of have to because, you know, like again, we talked about US foreign policy.

Um, Germany doesn't know if the US is going to be the same type of ally. Trump for years, even going back to his first administration, obviously encouraging the NATO countries to sort of more formally meet their commitments. So you have increased uh spending.

At the same time, you have tariffs which are going to make business more costly, less effic um probably less efficient because you're not going to have the most optimally designed global supply chain that would exist in a world of uh sort of pure free trade.

So you have this combination of every country sort of having to become a little bit more self-reliant at least in some respects. And you certainly see that in a lot of like tech areas, weapons, probably food even as well, energy and so forth.

Combined with the fact that, you know, you're having sort of less efficient supply chains, that's probably all things considered means um you know, more inflation and for uh to hit inflation targets, central banks are going to have to generally speaking keep rates up. We might get rate cuts in the US in the short term.

That's what uh Steven Englander thinks. That's certainly what the market is pricing. But he thinks the window for that will be pretty short and maybe even by next year we could be talking about rate hikes again. Yeah, I know. I know. We need that AGI. We need that AGI to bring down rates.

Uh how do you think the big tech companies have been changing their tune around how they talk about uh it used to be all about safety which you you know I think a lot of people have kind of just gotten sick of safety argument. Now it's all product and all of a sudden it's we're not moving fast enough.

You know Siri isn't good. Yeah. It's confusing. It's a little bit confusing because I mean first of all I agree the safety story feels very yesterday. This is such a business story.

The market to for better or worse and I don't really have a view on safety but the market for better or worse seems to be dictating to the companies. It's like no you got to race ahead as fast as you can. You do not have an alternative to this and so forth.

And you even see it, you know, obviously like when Deep Seek came out and it displayed its reasoning. Open AAI had to respond and it started displaying some of its reasoning models. So there's the sheer competitive aspect that like you just sort of got to do what you got to do for like the product side.

On the other hand, there are things that the tech companies are doing that do not sort of seem to me, oh, within a year they're going to build or within a couple of years, um, you know, they're going to have these brains that are thousands of times more intelligent than the uh, than a human.

I saw there was something about open air. I think about a social network. They hired a CEO for like product and revenue. There was that news today about buying a heart. You probably already talked in does not these sound like ventures.

These sound like investments that are about making money for business and do not sound like we are building something insane in a lab that is going to change the world. So it feels like there is this impulse among the big AI companies of various flavors to basically build good businesses and profitable things.

rather than this technology that you know is going to render human existence obsolete in some way or another. Yeah. Yeah. And it feels like open AI as a lab there there could come a time where open AAI could kind of uh uh what's the right phrase for it?

But but effectively like OpenAI will be in a position where they can be like yeah we're a consumer internet company. Yeah.

And there will be other there there will be other lab there'll be other labs that don't have the luxury of saying that and it'll be interesting if there's like this kind of divergence in I mean have to keep the AGI nar the steelman on the AGI narrative is that like would we have gotten this amazing consumer technology and enterprise technology and all this value without the crazy AGI narrative that drove nonprofit investment in highly unprofitable research and it's it's kind of like I feel like everyone makes the uh the comparison to nuclear but when I play back the story on nuclear it's like America incredible Oenheimer Manhattan project to develop bomb which I don't see really like a good thing like the bomb obviously it's just like it only destroys things and then it immediately becomes this this arms race and this equilibrium is just that you have to build a ton of them but none of them get used they create no real economic value right um but the good ending of nuclear would have been energy too cheap to meter nuclear power everywhere.

We didn't get that. And my hope is that we're in the situation where we have the panic scare around AGI as a weapon. But we don't go the nuclear path where oh, you can't possibly have Chachi be dangerous.

Well, I I agree with you and I think if you look at like I mean it is absolutely extraordinary the rate of growth that big temp big tech companies are growing from. I mean, this is what's sort of breaking the brains of everyone in the market, which is that everyone's been long big tech forever.

They're the biggest companies in the world, and they're still putting up some of the biggest earnings growth that you've that you can find anywhere in the market off of a gigantic base.

And I do think part of that story is got to be that CEOs really believe to and I don't know exactly what they believe and I don't know exactly what AGI is and they all have different definitions but they probably all believe it's like no like we're not going to slow down our spending on this stuff because we haven't achieved that thing that would that's out there that will be so special if it's achieved.

Well, I got to ask Joe. Would you rather would you rather own Costco at 59 earnings 59 Costco at 59 earnings or Google at 19 rotisser chickens? You'll never get me to make you know you'll never financial advice. You'll never get me to give fin why is every you tell me why is everyone so negative on Google these days?

What's the story? I don't I think it's products that are really hard to access like like they're incredible research lab but they but they haven't communicated very effectively around the product. So they introduce the product and it's in a research preview.

They have a paper, but in order to access it like chat GPT when there's a new feature, you go to chat.

com and it just says use this thing and and with Google it's like there's the Gemini app and then there's Gemini smart responses and then but that's the product experience and then there's obviously like the the threat to the core business model wind up searching in ChachiBT you stop.

They got a kn they got a knife on the neck of the golden goose. Yeah. No, it's true. Um, the Google AI products, some of them are extraordinary. I've played around with some of the coding tools. I've notebook is a very useful tool for research. It is true that like no one would ever know how to find any of this stuff.

It's really hardc Well, it's in the app store now. Notebook LM's in the app store. It's in the It's notebook. What kind of name is that? Who would even know what that is? It does seem I don't know. What kind of name is Joe Weisenthal? Yeah, but what kind of name what kind of name is like 03 Mini?

Yeah, they're all bad names. The names The names won't The names won't matter. The names don't matter, I think. But yeah, there there's something about uh like Google is if they really really went full tilt into AI, they would just say Google. com is our AI product, right?

Instead of there's a separate app, Gemini, and like you can go over here and we have an extra thing instead of being like no like we are an AI company. Could you see that happening? Uh, no. No. I don't I don't think the company set up for it at all.

But I think that on a long time it might be different if it's if it's Mark Zuckerberg and he owns 100% of the shares and he can just do whatever he wants. It's private company and there's no investors that can sue. Like there are different structural dynamics at Google than at a small startup.

I can just see it being question. Yeah. Yeah. Please. Can I ask another question to you guys? Do you have a clear way to articulate the benefit that Meta is getting from all of its AI spending?

I mean, yes, because the first like the the the the base use of AI at at at Meta, the number one like just energy cost in GPUs is ads delivery. I know, but old fashioned machine learning algorithm stuff like what does that have to do with generative ads?

the same thing like like the the algorithm that decides what ad to serve you when you're scrolling Instagram is a big transformer now.

And so it's a bunch of weights just like an LLM, but instead of taking in text and putting out text, it takes in everything that you've looked at, how long you looked at it, uh what you liked about it, whether you better and better. And so that's a huge training run.

They never talk about this because it's not sexy and it's not the new hot thing. But but but you can think about like the the meta ads training run as similar to like a GPT3 GPT4 level training run. And then the reels algorithm for what content to serve you, not just what ads, that's also another huge vertical.

And so when Zuck went to build out the reals uh algorithm farm, the server farm that would power the reels algorithm because it's billions and billions of data points, right? Um, he said, "I don't want to get caught flat-footed again like I did against Tik Tok.

" Because when Tik Tok came up and they and the whole algorithm shifted to I need to rank every piece of content across the entire network and match billions and billions of of videos with billions and billions of users. He said, "I don't want to get caught flatfooted again.

Build me two data centers that are the same scale. " And then he had he had extra capacity thinking about, "Okay, I have two data centers because I just built double. Let's train an LLM. " And then the more and then the more tactical reason is that LLMs are going to get stuffed everywhere in every enterprise.

I've noticed and sometimes I just accidentally wander into an LL. Yeah. But not even from the consumer perspective.

Like think about think about within within Facebook they have to process payments from their from their advertisers and they might want to run those through an LLM to say are we being defrauded or or or there's a consumer uh there's customer support that's happening or any little internal thing even just like uh should should we uh should we censor this comment for a user who's under 18 years old right like that that is an LLM that runs in the background it's not consumer facing.

Why would they have to pay open AI for that? Why would they have to pay Google for that when they just makes a lot of sense? Can I say one more thing? Please. I I need a um for anyone who's watching, I have a lot of dear friends. Please never send me a real. I never watch them.

I like someone's like, "Oh, they send you a real. " It's like I'd rather not. So, I would like to use your platform to let Yeah. What's the best way to get my friends know I love them. Yeah. Send me a DM, whatever. I'm just not going to wa watch your reel. I'm going to I'm going to mail you a DVD of the best reels.

That's fine. I'm going to mail you a DVD, a Blu-ray of just all the best reels. Well, what about taking it a step further? Vinyl and it's us describing reels to Jeff. That'd be great. No, I love it. I would do that. We would do that all day. If someone sends me a DMs me a real, I'm like, do we really have to do that?

Imagine a bunch of colorful marbles falling down the stairs endlessly. Imagine subway surfers in the background while Joe Rogan talks about weightlifting with Elon Musk. Uh that's really high performance stuff, but you'll hear it on vinyl for sure. Thank you. I appreciate it. Anyway, this was fantastic.

Uh let's make this Let's make this regular. We This is such a fun conversation. Anytime. I love chatting with you. We're going to just put a calendar hold on. Yeah, we'll throw something.

If you ever just If you ever if any part of you ever thinks, I don't I'm not excited to talk to John and Jordy today, then we'll cancel. I'm always excited to talk. That's great. Yeah. Yeah. Well, we'd love to have you back. Thank you so much. I'll be back. Thank you for moving the market. This is great.

He moved the market on our stream numbers. We're up at 10. 8K today. Good day on X. The live stream is popping. Google's up. Google's up 5%. 8% today. 5% yesterday. That's because of us. Joe said, "Why? Why is everybody bearish? " You just had to ask that question and the stock popped almost 3%. Just reading tea leaves.

Hey chat GPTt, take this transcript from Joe Weisenthal and turn it into financial advice. Make it make it ultra opinionated. No. Uh obviously no financial advice on this show, especially for my guest. The question of like what does Zuck get from Llama?

We obviously did our Llama deep dive, but it's like such a history and and appreciation for open source and so many different platforms, right?

He just uh it would be hard to take meta seriously as yeah a hyperscaler as a hyperscaler if they didn't have any type of AI play they also get the benefit of learning from other people building on llama even if there's not some massive direct revenue benefit and and I think I think what Joe is is really like hitting on which is correct is that the LLM chatbased chatbot UI paradigm is not a perfect fit for what Meta's products are right now.

But as the as the UI paradigms mature, you can imagine just okay, we know that images are moving from diffusion to token based architectures.

So you could imagine that it maybe I don't want to constantly generate AI slop and throw that up on my story, but I could totally want to take a selfie and say, "Hey, uh, just with natural language, can you make the sky clearer? It's a little gray today. Can you just make it blue back there?

So, it's a hedge against as as as AI companions and just people develop relationships with effectively the models. That is a threat to Meta's core business of connecting regular people that have conversations in the app that that bring people back to the app so you know they can serve more ads. So, yeah.

Anyways, today was a very fun show. Very fun show. We got to go take some photos. We had range from Eric Prince to Joe Weisenthal. We really covered everyone. I got I got some text messages. Someone said, "Uh, great lineup today. " I'll read this up. Where is this? Uh, great lineup today. Blackwater.

Blackwater for software engineers. Cognition. Blackwater for Sam Alman's haters. Open AI. Uh, lots of Blackwater folks. Uh, the Blackwater of podcasting. Joe Wisenthal. Yeah. Or David Senra. I think David Center is more blackwater coded when it comes to how ruthless efficiency he pods. Anyway, thank you for listening.

Uh, this is fantastic. Uh, switch your business to ramp. com. Check out Poly Market talking about using ramp ramp. I want to get uh I'm going to get work on getting a Poly Market set up on PM Spack. Oh yeah. Like does it actually does it happen? When does it happen? When does it go out?

Like the IPO markets that we've been having a lot of fun with. And then as the rumors start circulating, who's the target? This is great stuff. I'm gonna be we're gonna hit up all our boys, all our group chats. Hey, who's got a profitable financial services company that they want to take public?

I want to I want to highlight one timeline post before we get out of here. Leonard Heim, he's coming on the show uh next week. Uh Jordan from China Talk recommended him. He talks about chips and AI. And I love this because there's a there's a screenshot that he's quote tweeting from, I believe, the Financial Times.

It says, "We still don't know how much a energy AI consumes. Companies must give us the chance to understand the environmental impact of the tech we use. " And he says, "Yes, we do. It's 21 gawatt.

You count all the AI chips produced, factor in that they're running most of the time, add some overhead, and you got your answer. It's a lot, and we'll only get more, but you know what? Probably worth it. " And I love I love that like there's so many of these opinion pieces that are like, "We could never know.

We could never know. " And he's like, "No, you can just add it up. It's pretty easy to understand how much energy it's producing. So, uh, I saw this post after I already invited him, but I'm I'm super excited to talk to him next week because, uh, clearly banger, banger. Anyway, it take Smith.

Leave us five stars on Apple Podcast, Spotify has been really explicit about this. If you like watching the show, our jobs are on the line. Our jobs on the line. They will recast us. They will cast someone else. Ben is making menace giving us a menacing look right now.

His his his thumb is over the purchase button on an Audi R8. And he can only do it if you leave us five stars. If you don't leave us five stars, we're never going to get there.