Tribe Talkin' Ep 126: How To Steal Tokens. Jail For Misleading Investors. DashDot Administration. Self-Interest In A Superhero Cape. Sharts.

Speaker A: Good afternoon. Good morning.

Speaker B: Morning, mate. Morning. My time.

Speaker A: Yeah, you sound like how I feel. What time is it there?

Speaker B: It's 4am Going really tired. This is the first time I've used my voice right now this morning.

Speaker A: So how's the travels?

Speaker B: Yeah, good, man. It's been more full on than I thought. The, uh, Marin's family has had a pretty intense schedule for us, which has been good, but next few days a bit more relaxed, which is awesome. But no, it's, it's beautiful over here. The weather's been amazing, the countryside is amazing. It's a very relaxed, peaceful existence over here.

Speaker A: Yeah, you sound practically shot and fucked, so I hope you do.

Speaker B: I'm still recovering post London. My sleep is still not great over here.

Speaker A: I don't know where Jack is.

Speaker B: Jackie boy. I'll be two minutes.

Speaker A: Okay. Well, there's a lot of. While we're waiting for Jack, have you been following the political. Like Carl Stickmanovic getting shunted from, um, Channel nine for interviewing, uh, a right wing lunatic and.

Speaker C: Oh, I actually watched it. Have you watched it, Don?

Speaker A: No.

Speaker C: So I watched it, um, when it first came out, fell asleep, got halfway through and then had to watch Pauline's update of it because she uploaded it to her own YouTube after it got deleted.

Speaker A: And what's the summary? Jack?

Speaker C: Like, he's not as, like the Tommy Robinson is sort of, I don't know, it's not. He's definitely done shit that deserves, um, like he deserves what he's got, but at the same time it's not any more than the people that are giving it back to him, if that makes sense. Like he's just a right wing activist and at the start of his journey he was fighting that with his fist, but now he's fighting it with his voice. But they're sort of characterizing him as a terrorist and I think that's, I think it's pretty extreme to the extent I've gone to.

Speaker A: Yeah, right. Uh, I do get the feeling just the sheer amount of social media clips that are quite extreme flying around at the moment. Surely Australia is in the middle of the grips of what the US previously had of unknown forces and energies wanting to push, uh, out the establishment.

Speaker C: It's a good, like, you should listen to it. Like, it's interesting because like, I think that's. Australia is definitely heading that way. Like UK is just in, in front of us. But with Kyle pushing out, I think it's going to happen way faster than people think.

Speaker B: Yeah.

Speaker A: Right.

Speaker B: I'll have to have a Listen, I only saw the headlines here, right. I'll have to go and get into it.

Speaker A: Yeah, I think so.

Speaker C: Yeah. But the screen, fuck, that was, that's inevitable. That was happening. Like I thought he was going to go out himself, but it looks like they've pushed him out before he could.

Speaker A: But what a way to go though.

Speaker C: Um, mate.

Speaker A: M. Talk about, you can't pay for this coverage.

Speaker C: Yeah, I didn't, I didn't realize how right wing he was. Aaron. I got the lineup of people who just been the most extreme right wing.

Speaker B: I'm not really on that much social media anymore, but definitely like Facebook is probably where I see it the most.

Speaker C: A lot of mates have turned really right wing to the point where like they're just sharing all the uh, like Patriot shit on Instagram, but then even just seeing the reels and like seeing who likes the type of reel. Far out. Like there's so many people getting behind this right wing Australia.

Speaker B: Interesting, Interesting. Apparently the uh. So apparently there's a WhatsApp group or something for the, the Gold coast startup community and apparently it's been blowing up with all these right wing combos and people started leaving the group and all sorts. I started getting messages about it with people assuming I'm in there, I'm like, I'm not in the group. Apparently it's just gone off.

Speaker C: Yeah, that's one thing I do hate is when politics gets in the way of friendships.

Speaker A: Uh, no. Uh, all right, I'm back. So I guess we just start now. Well, let's get into it. Given Aaron's uh, got a big day ahead of him, I thought, you know, one of the big things in this last week is that the carve out for capital gains tax has been announced and there's like essentially your first mil, sorry your first 10 mil if you meet the criteria for one of these companies is going to be under the old regime. In my experience, there's going to be a lot of. Can you imagine the accounting and lawyers and the shit fight, just making sure that you are classified as this. And there's already some talk of, well, well, at the date of incorporation you don't actually have anything. So is there a technically a loophole that they can use to cover this off? But um, Tim Doyle got a guernsey to go to the Senate hearings. Now the biggest change in Australian taxation history, Australian taxation in the last, uh, I don't know, hundred years, uh, they're doing, they did with two days of uh, Senate hearings. But anyway, Tim was there and he was fair to say he stuck it right up to all of the people up, jumping up and down. Um, he said founders are the wrong martyrs in the CGT fight. Most of the backlash was self interest in a superhero cape. The reform's real casualties are always the employees who take risks and hold employee equity. 10 mil. The $10 million ceiling covers virtually every employee outcome in the country. It excludes the very largest founder windfall, which is precisely what the gains, which is precisely the gains that need least sheltering under the proposed startups. He goes on to say that rank the inputs to a great company by scarcity. The rarest is teams. Capital comes second and already enjoins generous concessions. Founders come last. Starting a company so irrational against the the salaried alternative that the tax treatment barely registers in the decision. The supply of people willing to do it has never been a constraint in this country. I think the point that he raises that I think is a similar point to, um, an earlier one that he made, is that without wanting to be too aggressive, I don't think there is a shortage of founders, but there is definitely a shortage of complementary teams and a shortage of people that are willing to roll their sleeves up underneath visionary founders that have the competency, the capability to actually help build the vision. Ah. And I think, uh, that is a very good point, is it not?

Speaker B: I think so. I can't imagine a quality founder saying I didn't start because of a tax situation or because of a likely tax outcome. And we see. Well, I see founders who are talking about complex structuring and setting up overseas entities tend to be the ones like when they do that early tend to be the ones that don't go on to succeed. So. And I think he's right about the employee side. I mean this is something we were talking about with the esoc. Yeah, I actually like his comments.

Speaker A: What do you got, Jack?

Speaker C: I agree with Tim's comments. That's about all I got. I'm a bit over, uh, CCG and tax reforms. Let's just get back to building.

Speaker A: Well, I think one of the things that came out can't buy a property in Super M. So that's gone. And on the death of your spouse or if you get divorced, whoever keeps the property loses those concessions. So I have a feeling that the marriage rate within the cohort of people that are negative gearing, I have a feeling the divorce rate's about to fall, isn't it?

Speaker B: There's gonna be some weird outcomes for a little bit.

Speaker A: Unfortunately, I don't have a negative geared Property to be insurance to just extend my marriage a little bit, uh, a little bit further. Now like there's a bit of a spectrum. There's the people that go on to make the massive windfalls which Tim himself did. And some of the uh, flack that Tim has got from his comments is like, it's all well and good but you know, you got, you got your big windfall treated under the old, uh, old system. Then there's sort of the uh, the big part of the bell curve which is the founders that unfortunately give it their all but walk away with fuck all. And that's a uh, sad majority. And then there's a few bad eggs that at the other end get involved in a little bit of fraud. And so this is the first jailing I've seen in a long time. But this is an article in the afr. It says, Failed AI startup founder jailed for misleading investors. This is by Paul Smith. The former founder of failed artificial intelligence marketing company Medigy has been sentenced to nine years jail for misleading investors while capital raising and dishonestly using his position as a director to get multimillion dollar personal loan to spend on luxury homes. In November, David Fairfield admitted to one count of making false and misleading statements to investors who put in $15.7 million for shares during a uh, $50 million capital raising. 2021. And one count of dishonestly using the position as a director to gain financial advantage when borrowing 7.7 million in 2021. He used the money to buy a $10.5 million home in Sydney's Mossman and one worth 7.7 million in the Kangaroo Valley near the Southern Highlands on uh, anyway, blah, blah, blah, uh, he's got a non parole period of five years and four months. Um, you know, without wanting to get ourselves in trouble, there's some founders in some entities being pursued by some investors for fraudulent, uh, statements that I would have thought would be shitting themselves at uh, at this. But yeah, it happens every now and then. But I think it just has to be such a warning to founders that, yes, okay, in this instance it's so big and so crazy that clearly there's, I don't know, some mental health issues to deal with. But there are some times where people run gray areas thinking that they'll be able to solve it and um, and you get yourselves into trouble and it's just not worth it. Like why the fuck would you, what is going on for you that you needed to go and pull this off to get a $10.5 million house. Like I. Yeah, yeah.

Speaker B: This is pretty extreme. Like, uh, it's. That's quite a severe penalty. So this, this, um, like this should make some particular people very worried at the moment. Like you said, Don, Um, but I do think for founders there is a tendency, like first time founders, they sometimes make the mistake of treating the company almost like they're an extension of their personal income. Uh, and almost like an ATM and they blur the books and these things can get messy very fast. But when you have shareholders and when you've got investors putting capital in, this is a separate entity. It's not an extension of your personal wealth or your cash, to be fair.

Speaker A: I mean, those property prices only really get you a townhouse where Jack lives, like in Adelaide. Now, when you see the, when you see the value of property going through the roof, that's harsh. Yeah. And, uh, I think what there's always during boom times, people don't do checks. So like 20, 21. But the biggest risk is often where founders have been running a company for a long time and then raise capital. And there just can be sometimes this. What's the word? It's not expectation, it's. It's where you expect. Help me out here. My brain is. You know, when you just entitles, you have an. You have this entitled view that I started this thing, I'm the one doing all the hard yards, you know, blah, blah, blah. And sometimes I've just seen it, um, blur and it's just not, it's just not worth it and you get yourself into trouble. So anyway, don't be a dick. Um, now another. Let's just stick to some sad, hard luck stories, but this one again has some things that I can't quite understand, but Dashdot went into administration, uh, with customers collectively owed over $10 million, which is huge. It's quite amazing. Now, where it's connected to us a little bit in our ecosystem is that venture debt provider Mighty Partners has emerged as one of the victims in the collapse. As stranded customers erode that collective 10 million. I'll put a link to this. This is a deep dive by Jack Derwin and it's massive. It's a very long, detailed document. In order for this amount of money to be owed to customers, there's only really two ways. The cost base was so extreme that the cash got burnt in that manner, which, that is nothing more than mismanagement or money was taken out of the company to fund other things. And that's nothing more than mismanagement. But yeah, it's collectively, when you add up everything, the entity has $749 in the bank and 700 creditors. This article says owed 16.57 million. Now, a good friend of the show, who I won't mention, who's in the venture credit space a couple of weeks prior, I had a chat to him earlier today. A couple of weeks prior, um, they were pitched an emergency R and D funding facility. And what was pitched to them was a story that didn't quite make sense. And luckily for him, um, he didn't really do anything about it. And next minute this has happened. So, um, yeah, you gotta keep an eye out.

Speaker B: Yeah, this is huge. 10. 10 million to customers for prepaid services and refunds. Staff 1 staff are due 1.1 million in employee entitlements. And the venture, and the venture debt was 1.5 million that's owed. Like, they just let it go right down to the wire. Like to have 700 bucks left in the bank account. It's not like they called it early and put up stumps like, this is, um, incredibly poorly managed from the looks of it.

Speaker A: Yeah, we'll have to see what comes out. But again, this is why you gotta do your dd. This was a little bit like this was a buyer agency platform that was using AI to allegedly do different things. So it's sort of, sort of in our world, but it was sort of not so much because it was just trying to take an existing world and, you know, apply new, uh, apply new stuff to it. So customers funds have been taken. But this next article I, uh, saw in Capital Brief as well, talking about being taken. You can just steal tokens. This is an article by Bronwyn Klune. A new wave of AI fraud reshaping the software business. Token theft has emerged as a real risk in the age of AI. And according to stripes Elemental and according to stripes Emily Sands, it could spell the death of freemium software models. Um, so it says AI token theft has eclipsed. Has eclipsed traditional credential harvesting to become one of the most urgent and least discussed cyber security threats facing businesses embedding AI into their products. That's the assessment of Strife's global head of data and AI, Emily Sands, who team has a front row seat to emerging fraud wave and the consequences it has behind beyond the consequences it has beyond surprise API bills, including the end of freemium as a business model. AI has destroyed the zero marginal cost economics that made SaaS fraud barely worth attempting or protecting against. All that the vendor risked losing was a small monthly fee. But that has now changed and every business embedding AI is about to inherit the consequences. I thought this was quite a timely one and fascinating because especially as more and more people are building their own stuff, I. E. I'm made a rule now on my Claude that I'm not allowed to start anything new until I have finished everything else. But then with some of the family office stuff and how I'm trying to get them up to speed, I can absolutely see how token fraud could absolutely happen. And if we don't have some very clear strategies around it, we could get done.

Speaker B: Yeah, I can see this so easily happening to so many, uh, the token theft. It's been some interesting discussions over here around tokens, AI tokens becoming uh, almost like a currency and by taking it to extremes of AI tokens becoming almost like. Do you remember that movie? I can't remember what it was. Where you have the time code on your arm and it's how many life credits you have left and you exchange it for every.

Speaker A: Yes. With, um, Jason. Not Jason, um, the NSync singer.

Speaker B: I know who you mean, but I can't. I can't even place it at the moment.

Speaker A: I uh, got done for drink driving. Jack, come on, help us out here.

Speaker B: Who's m. The last to know who we're talking about?

Speaker C: No chance.

Speaker A: NSync. Okay, there's his face. Justin Timberlake. JT.

Speaker B: Yeah, but imagine it, you know, if people been forget universal basic income, it'll be universal basic token. Anyway. That, that's a whole side conversation, but I can see this easily happening in a number of ways. I'm. I'm not sure though, but the death of freemium, I think it's obviously contextual and that's a broad statement.

Speaker A: I can see how wondering how it impact. Yeah, for those wondering how it occurs, the uh, Bronwyn says unlike stolen SaaS credentials which unlock the product tokens unlock infrastructure, they can be resold, exploited, or used to run an entirely separate business on someone else's bill. To date, the most visible form of the threat involves stolen API keys and credentials. A compromised open source AI proxy tool called Light LLM with over 95 million monthly downloads was recently engineered by attackers to capture uh, every API key that passed through it. But often the token fraud is harder to see and it's harder to stop. Tokens can be resold on marketplaces used for domain specific exploitation, such as mass generating music tracks, uploading them to streaming platforms with fake listeners, and then collecting the royalties, or deployed to run rapid businesses that Clone an AI product and sell it at a discount without bearing any of the infrastructure costs. Like this is, this is the new world of, of stuff. Every week there's something new to be on top of and there'll be lots

Speaker B: of small ways this happens too. Like you can imagine employees in a company building something for themselves as their side hustle. That's using all of the corporate APIs or corporate account word Filet.

Speaker C: I was going to say I could, I could definitely see that happening right now in, in the big, um, incumbent businesses.

Speaker A: Oh, have we lost Jaron?

Speaker B: No, I'm still here.

Speaker A: Jack was saying that he can definitely see that happening right now in the big businesses.

Speaker B: Yeah, totally. I, I, I, I suspect it's happening a lot. People building businesses.

Speaker A: Does it even have to be the big businesses? Like if we, like how would you track, if you've got a small development team, how would you track that one of your developers has got a little side hustle and is using your tokens to do that on the side?

Speaker B: You wouldn't be able to. And I was even conscious of this with building Mishi for the mission and making that accessible via Slack and realizing people could try and use it in all different ways and trying to make it as tight as possible so that it can't be used just for researching something else on the side. And that that'd only be a minor, uh, token theft. But I don't, and I haven't actually seen anyone pitching how to resolve for this. Like, where are the audit trackers for token usage?

Speaker A: Yeah, well, I saw somebody use Mishi again. Even though the mission's finished, Mishi got a call up.

Speaker B: I, uh, I've used it from time to time. It's handy.

Speaker A: Yeah, well, these are new things to, to deal with. But we're also like, we're, we're not in the, I think we're in the middle of the round where old businesses, old SAS systems of record, we're going through more of a filtering system of those that are just going to die off and those that are going to survive. And Intercom, which later rebranded to Fin, was sort of, uh, quite a darling for a while, then fell well behind. But that team managed to really get behind AI and turn FIN into, I think it's fair to say it was really one of the first properties SaaS applications to take AI with both hands and provide usable, uh, AI to end punters. Would that be a fair assessment from your side, Jack?

Speaker C: Yeah, I think so. Like they were definitely early on the

Speaker A: customer agent side yeah, and they've done a stunning turnaround of a company that, you know, was a bit buggered to now being acquired by, for $3.6 billion by Salesforce, um, which is going to close, uh, next year. Um, this post said Fin started as intercom 15 years ago. We changed our name to cap our transformation just weeks ago. We were a darling of the SAS era and invented so many of the patterns you see in software today. Nearly four years ago, in need of a reboot, we jumped onto only weeks old modern LLMs to create and define a new category that we know as customer agents today. And I take my hats off to them, but there are so many, uh, smaller companies out there that are really struggling with this and I can just see more and more next year. If you haven't been able to pull off what they've pulled off with the speed of which things are going and you're going to be in trouble.

Speaker B: I think it demonstrates the founder's ability just to jump on something new and actually build something worthwhile from it. These guys have done incredibly well at that over, what was it, 15 years ago? They started like, there's a demonstrated operational excellence here that is just repeated and repeated and repeated.

Speaker A: I think not giving up. Imagine, sort of they were at their peak in 21. Could you imagine what their vow would have been in 21? Then they're in the toilet and then they come back. To transcend all of that culturally is very impressive.

Speaker B: The other thing I love about that one, there's the post, like the photo of the four guys together in what looks like a bar. Right Back in the early days, when you see founders, I think this is what energizes us so much. When you see founders and you see the potential of what they're doing and they're giving everything to it. And looking back on those original photos, it's, um, yeah, bloody well done to them.

Speaker C: I think for us, you look at our portfolio and then even just a lot of the founders that we chat to, like, it's inevitable that at some point in your journey you're going to be flat or on a declining trend. And these guys did that at scale, but still managed to totally turn things around. But with the new product. And it goes back to all the stuff that we always try to hammer home. And I know it's harder to do when you're actually in the weeds, but that's like, if the strategy's not working, you've got to pivot and like, these guys totally pivoted and Shipped a new product and that was Finn and that ultimately was what Intercom turned into Fin and it took over the whole business. So I think if they can do that at scale, like there's no reason why you can't do that. Sub 10 mil. ARR. Uh, when you got way less resources much earlier on in the journey, you're still trying to find product market fit even if you've lost it.

Speaker A: Yeah, a hundred percent. Jack. What. I'm thinking it through a little bit though. All of this stuff is interesting and hopefully it motivates. I do sometimes get a little bit stuck in try to help founders traverse it. And if you haven't, do you think, if you haven't been able to have an idea and solve it by now, is there still time or is this, are we in the zombie phase? That if you, if you haven't by now.

Speaker B: Yeah, I think there's, I think, I think there's still time. But Jesus. That like the compression of time now, like the pace of change, it's. It's like I feel like a unit of time is no longer the same as what it was even six months ago, three years ago. So time is definitely running out and

Speaker A: so if it is still possible. What, you know, the thing I can't understand is how can we be of more value? I don't. What else could we be doing to help traverse and that. It's a bit like reading those business books like good to great and they've curve fitted all of the businesses to fit the few criteria that they found. But good fucking luck trying to apply any of that shit to yourself. It's difficult.

Speaker C: I still think a lot of it goes back to like uh, the ability to actually execute. And what I keep going back to is like with going to the gym or dieting, you didn't need the personal trainer or the dietitian. Like it's just calories in, calories out. But everyone like humans always go back to and it's for that accountability piece. So I still think there's the role in the VC and I think that's one part we can help identify the problems and bring urgency to solve it. But I still think even in AI, like we said, Don, like your internal rule now is I can't start a new project unless I finish the last one. Like humans are just so prone to doing that and execution is still the piece that we have to do.

Speaker A: Yeah. And the more I get into some of the apps, the little thing, the deeper you get into integrations, the deeper you have Getting into, you know, screwing around with your Azure, uh, platform to get your Ms. Client ID and your APIs and, and all of these things. As a non technical person, you end up at this wall of frustration. I think the good thing though is it does force you to be a little more focused with your functionality, a little more focused with some of that uh, some of that time allocation and to really get your prompt right. Because the problem is if you get your prompt wrong and it creates this stuff and then you don't know how to bloody use the thing like that is so useless. That is that, uh, yeah. Good. Well, theme of the week. Uh, Jack, what do you got?

Speaker C: Don't really have a theme. I. So I know we said in London my award was I look like the oldest 27 year old. Well, this week I feel like the oldest 27 year old. Like I've hurt my back. I don't know how I've done it. But then I just really aggravated at the gym yesterday and like I've been in so much pain. So my theme would be uh, yeah, I can't believe I'm saying, like my back is aching. I'm m not even 30, but it hurts.

Speaker B: Is it related to the game on the weekend?

Speaker C: Oh, I think, um, what happened was I, uh, dislocated my finger a while ago and then I was still going to gym and then it just had no grip strength. So I must have been like overcompensating on my lower back because I had this pretty coming to London. I just thought I was all good. And then the guy got paired with at the gym was just kept trying to up the way and I was like, all right, like I can't not say take the weight off. And like we just went as high as we could. And then like last night I was like, holy. Like I've really

Speaker A: uh, hurt myself today. Okay, so it's actually an ego based on.

Speaker C: Yeah, there's definitely an ego component. Anna's got no sympathy because she's like, you already knew you had a bad back before you went to the gym and you still did it.

Speaker A: Well, there comes a turning point in every human's life, Jack, where you have to have this lesson. So good thing that you've um, achieved it now. Mine was at the age of six, but anyway.

Speaker B: Oh, uh, far out. Yep. My current theme of the week is how bloody hard it is to when you're in a different time zone. Like I have a real appreciation re appreciation maybe for Aussie founders that are say in the UK or Us and they're juggling multiple time zones. It is amazing how fast it is a killer of, I don't know, just partly productivity, but just rhythm, workflow rhythm. And I'm sure you find a workflow rhythm with time. But hearing the founders in the UK talking about, oh yeah, we're up early for this, we have to jump on for these UK calls, uh, Australia calls. And then we tend to sometimes do calls at. In between. I just feeling that a bit more this week and it's like, yeah, that it is m. It is much more difficult than kind of the tokenism of a statement.

Speaker A: So, yeah, I guess mine is. I've just had a couple of interactions with some founders this week and, um, they're seeing through the noise. The ones that are doing all right are the ones that have just got the future vision and they're sort of, they're able to sort of see through the noise and it's keeping their decision making quite good. And I don't know if you've seen, but SpaceX is already down. Like, it's, it's down 26%. Is it? Let's have a look. Yeah, 26, nearly 27%. And then just watching a few emails and a few news things, just how quickly that sentiment has changed, like it's the same fucking company. And then now you see the, the things coming out about, you know, is this the, uh, time the bubble breaks and what if this happens? And it's just like, it's just we live in a world where there is so much noise and these couple of founders that are, that have got a lot of external stimulus upon them, but they're doing really well at just going, look, all of this is noise. I'm just gonna stick to the, stick to the long term. Whereas others that are just letting themselves get pushed around, they're getting pushed around by fears around competitors or some investor has said this, or, you know, blah, blah, blah. And you can just see the sheer volume of decisions and the stress and strain is just a lot more than people that just keep calm and carry on. And so as this world gets crazier, my, uh, suggestion to people is like, what are the tools and what are the frameworks that every morning when you wake up, you're doing the best you can to keep focused on that long term rather than just being flapping around in the wind. And I say that as somebody who's suffering a little bit of that and needs to get on top of it. All right, well, a few shots. So this is so just A reminder of some of the interesting stats that for Nvidia, the top three Nvidia customers are uh, 64% of their accounts receivable. So this is someone the size of Nvidia with 64% of it from three customers. And the Tosc, their top customer is still 30% which is up from 21% in 2020. And so there's, there's just some, this doesn't really mean all that much other than man, do we have some, uh, weight on a few names. Because imagine if one of those got themselves into a pickle. Just the domino effects that occur. So this is a chart of Brisbane property listings for houses, apartments and other. Have a look at Brisbane. Since the budget announcements like for this is just existing houses, apartments had a little bit of a pop, but it's actually, uh, it's actually houses people seem to be running for the hills, um, on which I find find interesting. And then coupled with that, this chart is the Brisbane weekly auction clearance rates. So this is starting in March and so from March through till May we fell from 56 to 48% clearance. So we were already like, we're already weakening. And then you can just see budget stuff gets announced and that clearance rate pulls from 48 down to 31, which in the context, you know, as so much of Australia's economy is built on our residential housing market. It's going to be very interesting to see how this shakes out. This is a chart going back from 2012. This is the value in billions of the weekly flows into semiconductors. And this, this looks like a Richter scale from a massive fucking earthquake. Like it's all little, little. And then in the last couple of years we've gone from like, like 500% increases on a weekly basis. Like this is really quite amazing. But when that pendulum turns, it turns hard. Um, so Nvidia now joins the wave of borrowing for AI infrastructure. So this is a chart showing Alphabet, Amazon, Meta, Oracle and Nvidia of their borrowings in different denominations. And Nvidia now has joined the party, raising 25 billion in debt. Oracle's at like 50, Meta's at 55, Amazon's over 80 billion. And Alphabet is uh, what's that, nearly 90 billion. So it's not just equity. Debt is coming into it. And every, uh, every time companies have gone a little bit too far, it's always been debt. One of the things we've often discussed, this is a chart showing the gap in performance between closed frontier models and open weight models, um, and open source models and One of the questions is where's the trillions of dollars going to accrue at the frontier model layer? And we have a bit of a view and at least a question that a commodity product subject to competition with no switching costs, the marginal profit often goes to zero. And so I have a personal belief that the frontier models are probably in a bit of a pickle. But this is a chart that just shows that back in 2023 it was about a 12 month lag of the ability of an open source model to do what the closed frontier model did. But that gap has now closed to like four months and they're both up and to the right. And the challenge that I see here is like where's the proprietary aspect for the LLMs and sorry for the frontier models? And I just, I do wonder like in the years to come for certain tasks, why the hell would you spend money on Claude Opus tokens for certain repetitive tasks? Aren't you just going to go to a uh, local open source model?

Speaker B: Yep.

Speaker A: This is a chart going back to 1988 that shows a range of different cycles and then what the retail investor's cash allocation was and extremely low level seems to be 14 doesn't have a grade of what that actually means, but we're at that extremely low level which normally correlates to a little bit of a market, a little bit of a market shutter. But it's time in the market rather than timing of the market. If you were too cute and decided to sell when the market was going down, you'd be uh, you'd be in a lot of trouble. So yeah, margin debt has now entered a level not seen since 2007. What happened after 2007? I think everything was totally fine, wasn't it? And it's about to approach the height of 2000. Those are two years that preceded uh, some pretty, pretty big stuff. And this is another one like margin debt expansion versus contraction. We're still, we're still in the expansion phase but uh, often once it peaks in the expansion phrase, once it starts coming down, um, yeah, uh, it doesn't bode well. Now this is a chart going back to 1990. So what this is is the trailing PE, the forward PE, the CAPE, the PB, the PE s, the EV to EBITDA quick ratio. It seems to be a average percentile, full history going back to 1900. And this does seem to have a little bit of a, a uh, cycle to it. But it peaked in 1929, 1965, the dot com bubble and now we are exceeding 1929. Maybe it's different this time. Who would know? We're about to find out. And I found this one interesting. This is a chart going back to April 2025 that for a lot of 2025 the Russell 2000 companies that had positive earnings per share and negative earnings per share were pretty, pretty together but sort of from September 2025 there's been these periods of time where actually the negative earning per share companies, the company's burning cash, have well and truly outperformed the positive earning per share companies. So that's going to be really interesting to see. And then finally. No, I'll skip that one. Can we zoom in on this? This is AI recording apps. This is growth since June 2025 and this is app downloads. So this is really interesting. Granola has grown nearly 400% but still is only actually quite small in terms of app downloads versus Notion and GoodNotes. They have a lot of app downloads but aren't growing as high. But I have a feeling that um, granola is just going to overtake, isn't it? It's. I kick myself now if I don't hang up the phone on certain calls and say I'm going to call you back so I can call them back from granola. Like it, it just, I think granola will eat the world. It's just a great product. And uh, those are uh, the shards. What are you looking forward to?

Speaker B: The rest of my time in Germany. It is pretty epic. Hanging out over here. Yeah, yeah. So good.

Speaker C: What have you got planned today?

Speaker B: Well, in uh, about 90 minutes we jump in the car and drive to Europa park which is a theme park over here. So that means that the kids can go off on roller coasters. And I've realized as I've gotten older my, my um, my fear of heights has become a thing. Like yesterday we went up the top of this cathedral, it's the highest cathedral in the world, apparently went up the top of the spire. And um, unfortunately it wasn't just me that had a bit of a issue with the height. So Samaran and Tucker were pretty freaked out by the time we got to the top of this tiny little spiral staircase. But um, yeah, as I'm getting older I'm just realizing, yeah the things that things I used to just not care about at all. And suddenly now an issue.

Speaker C: That's funny. I'm looking forward to uh, just having a chill down weekend. We've got mate's birthday on Sunday but it'll just be a bit of a Lunch. So it'll be low seaweed.

Speaker A: It'll have to be if you have to be lying down.

Speaker C: Yeah, yeah, I can walk. It's just got a real bad hidden.

Speaker A: Yeah, I was looking forward to some sleep. I, uh, I had asked the lovely nurse last night that the second. Because she thought Matilda could be let out at 2 p. At, uh, 2am and she. She said, would you like to stay? And I was like, no, no, no. The second Matilda can leave. We are, uh, out of here, please. She said, okay, no problem. But I've then managed to. I was still playing on my computer and say like 1:30. And I thought, I'll just close my eyes until 2, which was dangerous because then I wake up at 2:45. So I've then missed the window. And then she's like, well, you'll have to wait till 5 now in order to do the discharge things. And then I just stayed angry for, uh, a couple of hours, so I couldn't then get back to sleep. And then I was supposed to be having my coronary CT scan this morning. Um, but I've got something. I've got some kind of thing and it's. I think it's slightly triggered my asthma. And one of the risk factors with the dye is actually asthma because it can actually trigger a few things. And Vu, when he was telling me about it, he did say that, you know, people have died, but don't worry, the risk is low. But, uh, I've got a little bit too much on my plate to try and do it dead. So I decided to pull the pin and I'll do it next week.

Speaker B: Bloody hell, big fella. Yeah. Rest up.

Speaker A: I shall. All right, team. Will you enjoy? Jack, I hope you recover well. I'm hoping in Germany, though. Will the theme parks at least be efficient? Everyone will be in line, they'll run on time.

Speaker B: Yes. The turf efficiency. No, it should be really bloody good. I'm looking forward to it all.

Speaker A: Uh, right. See you guys.

Speaker C: See you guys, mate.

Speaker B: See, Yes.