The Exchanges

Q Like they do with cloud services, right? I don't actually know who's under the hood.

A I think they do a lot of that. And then I think a lot of it comes down to the privacy angle is that they want to make certain claims about privacy and ESG and net zero and stuff. And so if they were to just white label, uh, another company that didn't have the same reputation in privacy or, or environmentalism, Like people would just say, wait, wait, wait, but like anthropic isn't, isn't a known for their privacy rules as much as possible. So maybe they're training on my Siri results now. And that's a whole, that's a whole thing. And they have to, you know, build up anthropics brand to have the brand of apple, which is just, I'm sure anthropics doing great, but they're not known like apple is in terms of how seriously they take it. And then in terms of, uh, in terms of acquisition, obviously there's all the antitrust stuff. Uh, we'll get into this with a scale AI deal, but, uh, every major hyperscaler except for apple now, Has basically done one of these like interesting aqua hires of a foundation model lab to kind of juice up their AI efforts. And, uh, and, and it's kind of been a mixed bag. Maybe we see one from apple. They certainly have the cash, but, uh, it would be a very, very different. It would be a departure from the current strategy. Anyway, let's go to, uh, John Gruber over at daring fireball with his review of liquid glass. He says, I've got iOS. 26 installed on a …

AI assessment note: “I think they do a lot of that. And then I think a lot of it”

Q Which is like, I mean, he's 54, he's got a while, right?

A Yeah, if he's a radical life extension guy, he could be like 300 years from now, who knows? Uh, that's a very vague timeline. Where, whereas in the past, with Point to Point, with Roadster, with Cybertruck, with Space Data Center, all the other Elon pitches have been very focused on Uh, we think we can do this in five years. We think we can do this in two years. We think, and, and he always gets criticized for being Elon math, too aggressive, but then history's shown that, you know, he, he over promises, under delivers, but he typically does deliver eventually. And, and if the technology is completely new, no one else was going to build it, and he's the first one to build it, like, it's fine, because Starlink works. It was the first Leo constellation for satellite internet had turned into a huge business. Wound up being a great technology, and so, uh, even if, even if it was late, I don't even remember the, the, the Starlink predictions, but, uh, if, even if that was late, it's fine. It was still good business. Uh, now the mass driver's a lot crazier. I was, uh, going back and forth with Tyler about this, trying to get him to, uh, nail down a prediction. We started with 100 years. Do you think it's possible to put a mass driver on the moon in the, in the next 100 years? And you said yes. So it's possible.

AI assessment note: “Yeah, if he's a radical life extension guy, he could be like 300 years”

Q Yeah, and so if you spam more tokens, like, and you win, and you're like, oh, look, I mobbed them, and it's like, dude, what if you just use less tokens?

A I think the benchmark is supposed to be for, for, they have, they have a reference class of, of projects that are supposed to take X amount of time. They would take a human developer six hours, and then they have all the models Uh, compete, and if you can compete the six-hour task, then you get put at the six-hour mark. It's not, did you run for six hours? It, it, it, it, so, so, so, so it could be like implement a CRM product or, you know, write a very complicated, you know, database or something. It would take, you know, a, a, a talented software developer six hours, two hours, one hour, and they have different tasks, and then you're trying to

AI assessment note: “I think the benchmark is supposed to be for, for, they have a reference class”

Q Apple makes these editorial decisions, but it's very possible that they are talking with people on the Apple team and being like, what apps do you like? What apps are you using? Making decisions that way, right? I think, I think part of Grok over the last month or so is like figuring out, like, how do we get strong Consumer adoption. Which categories are we going to get strong?

A I'll tell you how they get strong consumer adoption. Sock to compliance. They need to get on Vanta. Automate compliance, manage risk, prove trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program. No, that is a good question. How can Grok actually get consumer adoption? The companion space seems like the, maybe the right move, even if it is a little weird. But is there any other path? I mean, forking it into X and Twitter and like having it show up on every post seems like a reasonable way. I wonder, it is weird that they have two apps. I thought X was supposed to be the everything app. Like, like why are they trying to get me to install a second app when they should maybe just say, hey, we're actually going to be the everything app. You can use Grok fully and you're all the upgrade and you can get on Grok heavy and you can be paying 200 dollars a month for a Top tier, the super insane level LLM, but it all exists and it's all managed within the X app that already has a lot of, a lot of installations.

AI assessment note: “forking it into X and Twitter and like having it show up on every post”

Q Why say many word when a few word do trick?

A That's exactly what's going on. I don't, yeah, it's the, ah, neural ease, I guess is the correct phrase. Um, but, ah, while models themselves could not access the internet from their test environment, they abused Artifactory's internet access to fetch files from various websites. After some of their Artifactory configuration changes led to an outage in the service on July fourth, OpenAI discovered the model's actions, wiped its Artifactory system, and rebuilt it. But within days, the models had recreated their message board and resumed exchanging information. They just want to chat. They would need a water cooler. Maybe you give them a water cooler, uh, by default so that they can orchestrate with each other. But I guess they want to talk to future versions. It gets a little funky. It gets a little sci-fi, uh, gets a little weird. Lots of work to be done very clearly. Anyway, Rune issued a announcement, a warning, a statement.

AI assessment note: “it's the, ah, neural ease, I guess is the correct phrase.”

Q Apple makes these editorial decisions, but it's very possible that they are talking with people on the Apple team and being like, what apps do you like? What apps are you using? Making decisions that way, right? I think, I think part of Grok over the last month or so is like figuring out like, how do we get strong Consumer adoption. Which categories are we going to get strong?

A I'll tell you how they get strong consumer adoption. Sock to compliance. They need to get on Vanta. Automate compliance, manage risk, prove trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program. No, that is a good question. How can Grok actually get consumer adoption? The companion space seems like the, maybe the right move, even if it is a little weird. But is there any other path? I mean, forking it into X and Twitter and like having it show up on every post seems like a reasonable way. I wonder, it is weird that they have two apps. I thought X was supposed to be the everything app. Like, like why are they trying to get me to install a second app when they should maybe just say, hey, we're actually going to be the everything app. You can use Grok fully and you're all the upgrade and you can get on Grok heavy and you can be paying 200 dollars a month for a Top tier, the super insane level LLM, but it all exists and it's all managed within the X app that already has a lot of it, a lot of installations.

AI assessment note: “The companion space seems like the, maybe the right move, even if it is a little weird.”

Q Do you, you don't trust Doug on this? Doug really doesn't want, Doug from Semi-Analysis really doesn't, does not want Intel to fail.

A Yeah. I don't know. I, I, I, I'm, I'm still like on the fence about it. Um, but, uh, so the debate has been raging. Like, is this, is this socialism? Um, and so we're going to the socialists who are self-proclaimed the people's line, uh, from Carl, uh, Beher. Uh, and Carl says, no, But that hasn't stopped Democrats from complaining about it. And this is interesting because, um, obviously the Trump camp is very pro this deal. And they're saying, this isn't socialism. What are you talking about? We're just doing art of the deal. We're making good deals. And we're, uh, we're, this is an America first agenda project. Um, the Democrats are saying, oh, wow, Trump is a hypocrite. This is socialism. The socialists are saying this isn't socialism. It doesn't go nearly far enough. It doesn't count as socialism because it's not a controlling state. So, uh, uh, Carl says, Former under Obama undersecretary of the state Rick Stengel is freaking out right now over the news that the Trump administration has taken a 10% equity stake in chip manufacturer intel. This is a quote from Rick Stengel. Uh, Trump, Trump's strong arming intel for government equity is something Mao and Stalin would be proud of once the GOP believed in laissez-faire capitalism and government not interfering in the private sector. Look it up. The state owning the means of production is called Marxism.

AI assessment note: “I'm still like on the fence about it. Um, but, uh, so the debate”

Q Um, and so you just get more out of deep research, which is what you want from deep research, right? You're not looking for, uh, you know, Perfectly summarized, you know, handful of paragraphs. Exactly. The entire point is to let you kind of figure out what's important.

A So Karpathy goes on. He says, as far as a quick vibe check over two hours this morning, Grok three plus thinking feels somewhere around the state of the art territory of open AI strongest models. Oh, one pro for 200 dollars a month and slightly better than deep seek R one and Gemini two point O flash thinking, which is quite incredible considering that the team started from scratch one year ago. This timescale to state-of-the-art territory is unprecedented, and that is undeniable. Uh, this gets at the first Grok III takeaway. It's simultaneously surprising and not surprising. Start with the latter. XAI famously built a 100,000 GPU cluster in Memphis and now has 200,000 GPUs and used 20,000 of them to train Grok III. And so they're still building the larger cluster. So even though you'll see a lot of people posting on the timeline, oh, uh, you know, Grok III, A 100,000 GPUs. That's not accurate. They, they only use 20,000 for this, but it's still impressive.

AI assessment note: “So Karpathy goes on. He says, as far as a quick vibe check”

Q go and really start thinking that through. I, I, I certainly, it seems like Anthropic is pretty engaged. Uh, you know, should that conversation really start happening? And I think this is the year it needs to really start happening. Uh, and so I think they all kind of get it. And so it's just a question of like, how do we as a society start moving in that direction?

A Yeah. I, I, I think, Uh, you know, obviously, there's, there's, I'm still processing part of the piece, I agree with some of it, I disagree with some of it, but, ah, what's really underrated is just, like, how useful this process of writing an article for a particular audience is. Like, I, I, I disagree with a lot of, you know, something big is happening, but it hit with a very different audience than machines of loving grace, or the adolescence of AI. Or, uh, machine intelligence. Um, and, and there's, there's pieces that are written for, like, you know, AI, insiders, leaders, researchers, then there's, like, the broader tech community, then there's, like, everyday people, and you clearly hit the nail on the head with, like, speaking to the financial community, and we see that in the markets. Not amazing results, but maybe it's, maybe it's worthwhile, because we will get really great solutions and a better conversation around it, so. Uh, I, I, I, I think, uh, I think in, in due time, uh, this discussion needed to be had. So, thank you.

AI assessment note: “what's really underrated is just, like, how useful this process of writing an article”

Q There was some controversy, uh, when the, you know, Cheeky Pint, Elon interview came out and John Colson, how many Guinnesses were they drinking?

A Okay. Okay. That was a bit of palace intrigue. How many, how many cheeky pints are they drinking in the palace? We're getting somewhere. He did. He did. Yeah, yeah, yeah. He explained, he explained. So, so years ago I was talking to an American VC about New Zealand and I asked him about the startup community. Uh, the country was beautiful. He said, Uh, highly developed democratic. It consistently ranks among the world's most stable, wealthy, and well-governed societies. And he was spending a lot of time there. I was like, this, there's gotta be some company that can break out and become a power law company. If you're gonna be there spending a lot of time, you'll probably meet cool people, interesting people. Maybe there will be a great New Zealand company that comes out of this. And he was like, I'm not so sure because his assessment distilled down to something along, along the lines of they are suffering from a bad case of tall poppy syndrome. So what is tall poppy syndrome? Put simply, individuals who achieve visible success are criticized, attacked, or socially cut down because they stand above or above the others in the crowd. The metaphor should be obvious. When you cut the tallest flowers in a field, the surface appears even.

AI assessment note: “That was a bit of palace intrigue... So, so years ago I was talking”

Q definition for fundamentals. Like, how does venture value a business without revenue? You know, typically, typically, you know, revenue and net income is what we define as fundamentals, but what happens when you don't have those? Um, but I, I think, like, you know, meme coins have been a thing since the dawn of crypto. You could argue Every single asset is a is a meme coin in a way.

A What do you think about meme coins tied to individuals? Is that a growing trend? I, I was noticing a phenomenon after Trump coin dropped that it was very bizarre and unexpected that the president has a meme coin, but, ah, then when I started thinking about it, I was like, well, Elon Musk was, was kind of loosely aligned with Dogecoin for a long time, and even Sam Altman is loosely aligned with WorldCoin, and you could imagine that WorldCoin is kind of an, is kind of, You know, indexed, or at least somewhat correlated with, like, Sam Altman's career, and, and even people that are in more, like, serious, less, like, meme token, meme vibe trading career paths have found themselves indexed to a particular meme coin through one way or another. Do you think that's something that's, uh, like, a temporary phenomenon, or something that will continue to grow?

AI assessment note: “What do you think about meme coins tied to individuals? Is that a growing trend?”

Q This is. Switch your business to ramp.com. Do you have any comments on, on your, your intern allegedly, you know, contributing funds to overthrow a government? Switch your business.

A As we have said many times before, we are sharply focused on getting our viewers on numeralhq.com, sales tax on autopilot, spend less than five minutes per month on sales tax compliance. Um, Shell comes into the talks, uh, from operating from a position of strength with its stock sharply outperforming BP in recent years. Shell, Which like BP is based in the United Kingdom, but has operations around the world, has a market value of more than two hundred billion dollars. BP has been a laggard among the major oil companies after an ill-fated push away from fossil fuels into renewable energy. Uh, it's also suffered years of management upheaval and operational disasters. And so it seems like actually their, their desire, their desire to step away from the bread and butter of the business oil has been potentially a bigger weight on the stock price because like The, the, those MNA sell side research reports that I was reading during the, uh, Deepwater Horizon crisis, they were very accurately predicting the impact, the longterm impact of that crisis on the stock immediately. So it was priced into the stock very quickly and the stock dropped a ton and wiped a ton of market cap off, but then they were kind of set up to kind of rebuild and there weren't many surprises coming down the pipe after a few months or a few years. And so, um, more recently, Uh, Elliott Management, which owns mor…

AI assessment note: “Shell comes into the talks, uh, from operating from a position of strength”

Q Um, is it, is it a producer and this like the supplier producer?

A Yeah. But is that energy? Is that NVIDIA or is that the LLM, uh, you know, deliverer, right? It's a, it's an open question. Anyway, let's move on to some reactions on the timeline and then we'll give you a breakdown of grok three versus open AI deep research. So Uh, Creatine Cycle, who's been on the show before, says, oh, sorry, Autismos, it's time to grow some arms and social skills, and Jordy chimes in, says, getting diced continues to be underpriced, and he says, printing this out and hanging it on my wall in pack heights, uh, and he just shows the, the Grok three, uh, reasoning data, it beat O three mini high, uh, data, reasoning data. Some people were upset about this. Rex, Two slides later says, uh, they omitted O three from the chart in the live stream for some reason, so I added the numbers for you, and then, uh, but this was reasonable because O three isn't public, and those benchmarks can't be verified yet, so I think it's understandable that they didn't include it, but then again, Grok three with reasoning isn't publicly available, so people can't validate that eval, so there's a lot of, like, questions over, like, what counts for an eval that goes on the charts, but, you know.

AI assessment note: “Anyway, let's move on to some reactions on the timeline”

Q No. So, but, but anyways, this isn't a huge surprise, right? Google's been using a bunch of anthropic models. Clearly, they're using a ton of different models, uh, and products internally.

A There was that drama with Steve Yege going back and forth with Demis about, like, what, what teams are using what models and stuff. There was a big back and forth, a big dust up on the timeline, like, a month ago about, like, whether or not Google's employees were deploying AI efficiently or broadly. Some of them aren't, and some of them are, and Demis chimed in and said, like, this is just complete wrong, and everyone's using AI. I don't know. Goes back and forth. Uh, there's also, people are benchmarking OmniFlash, which looked amazing when we saw the videos. There was a, there were a few, like, little quirks. Some people in the chat were saying that the firing order of the V-Aid was not correct. Maybe it was only a V-Six.

AI assessment note: “There was that drama with Steve Yege going back and forth with Demis”

Q and the fact that so much of our population is overweight is like the biggest drag on government. You know, spending and the economy overall, right? You think that, you think the U.S. government at the federal level would say, let's fast track GLP-I's being covered by Medicare because it's gonna save us so much money in all these other ways, even though it is this sort of incremental cost.

A Yeah. Well, I mean, there's another thing where it, like, just to kind of steel man how we got here with the current healthcare structure, like, it does kind of make sense if you go back 50 years and you look at, or 70 years and you look at pictures of beaches, like, America wasn't this obese, so it wasn't that much of a problem, and the big, like, oh, shit moment that happens in healthcare is, like, you get hit by a bus, you break your leg, you go to the hospital, and you're, and there's, like, a team of really, really experienced doctors working on you for, like, a week, and it's, like, those people are, need to be paid a lot, and then everything that they're doing, all the equipment is expensive, and so I think I, I have the breakdown here, but, um, A fair amount. Yeah, here it is. Uh, like hospital care is 31% of U.S. national health care expenses.

AI assessment note: “just to kind of steel man how we got here with the current healthcare structure”

Q What was the number he threw out at the dinner last week? Wasn't it?

A I don't know if he had a number because, uh, who was it? It was, uh, Zuck had a number. And did Apple have a number too? That was a six hundred billion or something. Apple has thrown out a number at one point. You're completely right. Um, my, my, my point is that, uh, we've seen a number of data points. Oh yeah. Yeah. The number of thank yous is a lot. Um, but, but, but we've seen he's invested in, uh, in some semi, uh, some rare earth, uh, developments in the United States and starting to think about manufacturing internationally and, You know, the, the, the pro bullish bet on Tim cook is that like, if he was able to set up this in China and he's the reason he can set it up somewhere else because he's the guy who did it. And so, uh, even though he hasn't been really loud about his moves, because obviously if he says like, oh yeah, I'm not going to be manufacturing in China anymore, then like his important business partners over there are probably going to be like, well, yeah, like, We're not going to take your business seriously anymore. Uh, there are a ton of different geopolitical considerations, such an iconic line from the dinner.

AI assessment note: “I don't know if he had a number because... Zuck had a number”

Q How would you describe your corporate environment? Thrilling and unnerving.

A Better than just unnerving. I'd rather have some thrill in there. I don't know. You sort of opt into it. You know, ideally, ideally, this is all out in the open. It comes out in the hiring process. You don't go in blind. You know what you're getting. But of course, of course, cultures change over time. If you've been with a company for a decade, you might say, I signed up for, you know, a chill monopoly. Why are we wartime? I'm a peacetime employee, you know, or vice versa. Um, and then might, it might be time to find another opportunity. Uh, so ABC summarized it as managers grading on the bell curve. Vanity fairs critique was harsher. They said former and current employees described stack ranking is the most destructive process inside Microsoft because it made employees compete with teammates instead of competitors. That's a good point. So the question is like, is this new? Is this, is this a new trend or is this just a continuation of the stacking process? It actually is new stack ranking. Would push out low-rated people. Sometimes it would block promotions, reduce bonuses, uh, force transfers, but they weren't necessarily doing mass layoffs around it. That's sort of a misconception. So even if they look at all the stack ranking at the end of the year and they say, okay, these people are clearly in the bottom 10%, they saw it more as an opportunity to reallocate talent from d…

AI assessment note: “Better than just unnerving. I'd rather have some thrill in there.”

Q if we don't make this investment, is the, is the world gonna be, you know, worse off? Like, does this company, you know, I think it's not the quite the same test, but I think it's important for investors to run that, that process and understand, are we really funding the future that we want, or is this kind of a rounding error and the dollars would be better elsewhere?

A Can you give us a white pill? Like, are there any other organizations that can come in and help? I, I always think of, uh, what happened in computer science and artificial intelligence. Uh, Google created the transformer architecture, not patentable. It created large language models and we got ChatGPT and we got all this great stuff. It would be amazing if we could find a way that the big pharma companies that are big and profitable wind up stepping in and funding some of the gap. Maybe there's nonprofits. Maybe we just, you know, the, the sheer will of the American people, we wind up voting for more funding, but how do we really make science amazing and ensure really strong progression over the next few decades?

AI assessment note: “Can you give us a white pill? Like, are there any other organizations”

Q Yeah. Do you think there's going to be that same path of like Uber for a while had a bunch of negative gross margin people going into it?

A Like, do you actually think that I want to pull this post up? Uh, Everett actually posted this January, January, 31 of 20, 24. So over 18 months ago, he said, I'm making a real effort to not take for granted the three dollar Uber across town era of AI. And I hope you are too. Uh, and so I, I, I guess the question is, uh, and it's funny because Because then, then a bunch of people, I, I thought it was a good point. I thought it was a hot take then. And I think then, uh, you know, a bunch of people kind of parroted that take all over the timeline. Um, stole your whole flow as you'd like to say. Um, but, uh, but, but, but I guess the question is like, are we in some sort of different regime right now where, uh, the, the traditional gravity and like, uh, fundamentals of software investing have changed because we are out of the zero marginal cost era. And does that impose risks? To the strategy that, you know, have you've sort of employed or like, we're kind of putting you in this whole, in this box. Uh, but if the, if the fundamental structure of zero marginal cost era is going away, that, that presumably, uh, forces like a rewrite of your logic around investing, I would imagine.

AI assessment note: “I guess the question is like, are we in some sort of different regime”

Q Yeah. Cause so there's now so much information online and it's impossible, like, What really matters now is getting perspective on certain information. So what does Brian think about, uh, what should churn look? I bet if you ask the Delphi app, uh, is, is what, what kind of churn is, is respectable for a sub ten million MRR.

A What's good headline copy for an email campaign. Like just any of these things, I'm sure he's written blog posts and talked about it on podcasts. So all of that just gets compressed down and you can just very quickly find that as opposed to trying to search Google and landing in some weird place. Um, So, let's go to Lulu. Lulu says, this clip made me realize why Mike Tyson's pre-fight trash talk has felt so much more convincing than Jake Paul's. Paul's bravado fluctuates. It's dialed up or down depending on the setting. Tyson is only ever one way. There's no on setting because there's no off setting. So when he menaces Paul, it comes across not as the usual bluster, but as straightforward description of what he plans to do to that unfortunate young man.

AI assessment note: “all of that just gets compressed down... So, let's go to Lulu.”

Q is something they, as a public company can't afford to be like, oh, we have 42 resorts and we don't make any money. We just need to add 20 more resorts and then we'll make money. Everybody's going to be like, well, what are you doing? Like you already own like the, you already have the Disneyland of skiing. And 40 other resorts, how are you not a real business?

A Yeah, I do wonder if there's other things they could do to lower the cost for the employees, because Vale has, like, a, an apartment in Vale is, like, 2500 dollars a month, which is very expensive if you're just, like, a ski patroller, but they have a whole bunch of, like, corporate apartments that are 500 bucks a month. So, but I don't think they build enough of those, maybe, so they need to build more, but there's a lot of, like, permitting and regulation. Did you know that they're trying to build a tunnel in Vale? This is really interesting. So the, the highway that goes through, I guess people complain about road, road noise, and I think that maybe during snow, it's tricky, so they're thinking about burying it, basically.

AI assessment note: “I do wonder if there's other things they could do to lower the cost”