Housing Crash Gets Real, NVIDIA’s AI Factory Gamble, Canva’s Markdown

0:00I'm Adam Schwab. I'm Adir Shifflin. And this is The Contrarians with Adam and Adir.

0:09And we are back with our super popular news episode on Saturday. It's been raining. It sucks off. So I've come back for another episode. So a lot's happening, Adir. I think, I thought let's go straight into housing and the housing crash. I've been calling it a housing crash for 20 years. And finally, the clock has struck 12. And recent data from Cotality reveals that Sydney house prices are dropping at a 16 % annualised rate over the past three months, while Melbourne, off obviously a horrendously low base, is off 13%. Even Perth and Brisbane are falling. As Chris Joy wrote last week in the AFR, if that pace persists for 12 months, it will be the single worst correction in Aussie home values in 46 years of cotality data.

0:51We already have the mother of all housing corrections on our hands. At the same time, Australia has one the highest rates of inflation in the developed world with headline inflation at 3.8 % and trim mean inflation at 3.6%. Bear in mind the US, which is currently encountering frenzied calls to raise interest rates, has core inflation of 2.4 % and headline of 3.4%. Well, so we've got the potential horror show of sticky inflation and terrible productivity growth in Australia, coupled with a generational property crash. Warren Hogan, of course, the former RBA governor, or former RBA member, I should say, called for the RBA to raise rates a couple of weeks ago saying an increase in rates would have boosted perception of independence, enhanced the board's inflation-providing credibility and sent a message to professional markets to lift their game on pricing term structure of interest rates.

1:38Adir, this is a bit of a mess, obviously caused by the federal government's budget from hell. What do you think is going to happen here? When that budget was announced, but before it was passed, there were a couple of main things that we said about it. Number one, we said this is going to tank house prices. That wasn't very dramatic to say that. It was pretty obvious. Number two, I think we were very early in saying this is going to be bad news for renters because everyone was saying the opposite at that point in time, or most people were. And it was very obvious to us for mechanics that we discussed at the time that this was going to be bad news for renters.

2:14And so putting that second one aside for the time being, I think this has played out exactly the way we thought it would play out. like you're basically removing the incentive for people to purchase homes that are not their primary residence and you might say well there should have been that never been that incentive in the first place and i'll largely agree with you about a lot of that but the thing is a minute ago there was that incentive and now there's not and you know people are just anchored to like well things are bad now and they used to be good we don't want to do that anymore like it's all this kind of sadness over.

2:49We could have had that, but now we can only have this. So I think that isn't going to turn around quickly. And then we say, well, we're waiting for all of the extra expenditure and price increase to come through in primary residence, because that's a very good place to invest your money. I mean, if you're happy to ride the housing boom, but I think that hasn't happened yet. Because the thing is this, if a property market is collapsing because all of the heat has come out of it due to, number one, the lack of incentive to buy non-primary residents. Number two, just the softening general economy, which is what we're feeling as well.

3:27People are reticent to start paying big money on their primary residents as well. I think that will eventually happen. I think there will be price inflation on primary residents as a tax strategy. But I think we're in this hole for a little while and I think things will get worse before they get better. And I don't think there's any obvious way that this can be fixed. And I think that when you hear, I think even Albanese came out and said, yeah, I'm happy about the house price crash. This is what we wanted. Now, I'm not sure that is true at all, that he wanted a house price crash. But when he says things like that, what it means is that people have said to him, this fall in house prices is not going to be turning around anytime soon.

4:09So you better embrace it and make the desire for it part of your narrative so that you're a winner when it keeps falling. And I think that that tells you everything you need to know about what the government thinks about this house price situation. Well, they're on the record as saying they wanted house prices to grow at a much slower rate. They definitely didn't want to crash. They didn't expect a crash. Chalmers said that. Albanese said that. So they can't rewrite that narrative because they were quoted. Of course they can. They're rewriting it. They're literally rewriting it. You understand how, what's it called that ministry in 1984 where he works that erases newspapers in the past and rewrites?

4:46Ministry of Truth. Ministry of Truth. You didn't read 1984? No. I've read Animal Farm. All right. Well, this is officially the last episode of this podcast. We're shaking his head in disgust. That is disgraceful. That is disgraceful. Have you read Animal Farm? Yeah, I love Animal Farm. He read Animal Farm. He said, yeah. Yeah, he said he read Animal Farm. Have you read 1984, Mike? I have read 1984, but I think Brave New World is an even better version of that concept. Oh, there we go. A bunch of buddy, a bunch of nerds. I think both are accurate, like have successfully predicted the time we live in now, but I think Brave New World is a more interesting version.

5:30Well, we can have long, I don't disagree with you, but I think, Adam, you need to go out and read this book immediately because it's totally ridiculous. No, I'm not reading it. I can't hate these sort of sci-fi futuristics. Obviously, it's written a long time ago, but not. It's not sci-fi and futuristic. Firstly, it can't be futuristic if it's literally set in 1984, although it was futuristic in 1948 when it was written. Yes, I agree with you on that point. Anyway, we're going to have this conversation offline, but if you would have read that book, and soon you will have read it, But you can say that he works for this thing called the Ministry of Truth, where everything that is called something, it's because it's the opposite, right?

6:12So the Ministry of Truth, their job is to manufacture lies. And what they do is go back and change historical newspapers so that the past matches their view of the present. And then they can control the future, basically, is the narrative of the book. And so basically, I think that that is very much what politicians are trying to do in this era. And I think that Albanese turning around and saying, yeah, we're happy that there's a property price pressure. That's good. Cheaper homes for first home buyers. Well, it turns out the one group that's not cheaper for is first home buyers, as it happens.

6:45But that is an effort to rewrite history. I think they're actively trying to do that because this has not gone the way they had expected. It has gone the way that other people expected it to go. 28 ,000 people who bought that absurd 5 % deposit thing, they're almost basically are underwater now. So some of these people might be losing their houses in the next year. Negative equity. Absolute unmitigated disaster. Can you see the comment from one of the ministers recently? Happily for them, I can't remember who it was. But like whoever the minister was basically said, well, these people can't have negative equity if they're paying off their loan.

7:19So they misunderstood servicing a mortgage for the mortgage being worth more than the property. Like, I just thought that was the embodiment of the cluelessness of the people that are in charge of running the show. It was just astonishing. I think the bigger issue, there's two underlying massive issues here that I've tried to paper over with this budget from hell. One is we've got an aging population that we've talked about that someone's got to pay for the healthcare of everybody who's turning 75 and nobody wants to pay. and everybody's just pretending it doesn't exist. So that's problem number one.

7:53And problem number two is the massively increased size of the public service, especially under these Labor governments. But Liberals are no panacea here, but certainly under Albanese and Chalmers, the federal government, look at the Victorian public service, 50 ,000 people, they've massively expanded healthier costs and massively expanded the public service, who of course vote for them, public service replacement theory. And this is why the budget simply cannot balance. It's leading to incredibly low productivity and this super sticky inflation. So as soon as you drop rates, inflation peaks up.

8:19And it's the only way to resolve this is to massively reduce spending. And they just simply refuse to do it. Well, I'll tell you one interesting thing about this house price fall. So I don't think we can say this yet. But if this house price fall continues for the next year, then it does say an interesting thing with respect to the argument that immigration is what's driving up house prices. because immigration will be largely unchanged, but house prices will be going down. And your argument has always been it's debt that's driving house prices more than anything else. And it will be interesting to see if that's true because that's going to be a tough pitch to blame it on immigration if immigration doesn't change and house prices continue to fall for a year.

9:02So far, you've seen loan applications to Westpac, CBA, etc., down 18 % to 24%. And you've seen house prices drop or particularly drop annualized. So, so far the debt theory is proven right. It's early days still, but that's 100 % what's happening thus far. I don't think it's bad. I just want to be clear about that. Like I don't think this fall in house prices is bad. Like I think house prices were out of control and the single biggest threat to Australia today as a country, in my view, is the unaffordability of housing for young people in a society where we still consider that the Australian dream includes owning your own home.

9:40So I do think that home affordability is the key issue. But I think I always say revolution, revolution is bad. Go read about any revolution and never goes well. Maybe, okay, you know, what is the Velvet Revolution in one of the Eastern European countries, right? That went okay. I can't remember which. It could have been Romania with Charchesky. No, that went badly as well, actually. That wasn't that one. I forgot we talked about Velvet Revolution. But generally what happens in a revolution is there's blood in the streets and a lot of innocent dead people. And it's much better to evolve to a happy place.

10:12And I think the problem we're seeing now is this is such a radical budget that it's driving a revolution in the economy and in property or the beginnings of a revolution. And that's what I fear. The change is too fast. Let's move on. And NVIDIA last week announced it would create an independent compute financing platform with some massive names in financing, Apollo, BlackRock, Blackstone, Brookfield, Goldman's, KKR to mobilize US$500 billion in third-party capital to help with the build-out of AI infrastructure over time, mostly data centers to produce, as well as fabs to produce the chips. Jensen Huang, of course, the founder, CEO of Amitya, claimed the move will create a new class of productive, investable infrastructure, AI factories.

10:54AI critic Ed Zitron wasn't so praiseworthy, claiming the announcement was merely an advertisement to customers and concerned investors about a non-existent fund. Ben Thompson, writing Stratechery, was equally sanguine, noting that it's not a total free ride for NVIDIA. The company is backstopping opportunities up to 25 % residual value-based finance, suggesting that Huang believes his investable asset class pitch more than the market does. That is, in a certain sense, a price cut as the goal to reduce the cost of capital for entities building data centers with NVIDIA chips, putting NVIDIA profits on the line for uncertain investments.

11:28The rationale may lay even deeper than the circular finances funding the entire bubble. Rather, Tom suggested it's a desperate move by NVIDIA to maintain its most powerful asset, which of course is CUDA, its programming language. Thompson noted that the use cases for those GPUs is very clear, and those use cases are happening at a much higher level than the CUDA frameworks, i.e. on top of the models. That is, combined with the massive incentives towards finding cheaper alternatives to NVIDIA, the pressure and the possibility of escaping CUDA is higher than it's ever been. Anthropic has not been dependent on CUDA for years and OpenAI is moving in that direction, at least for inference.

12:01If these companies win, then NVIDIA's profits will be squeezed. Andy, what do you make of all this? I think NVIDIA is the most expensive business in the world, 5.4 trillion. CUDA is really the gel that it's the switching costs and the cornered resource of NVIDIA. How much is this about protecting CUDA versus just keeping this Ponzi scheme, circular jerk, circular finance thing going well you say it's the most the biggest switching cost and in a sense you're right about that but well the reason you're right is nvidia's chips are the best chips it's just you don't really need those chips for most of the stuff that's being done so it's not like someone i think it's going to be a while till anyone really competes with the quality of the highest end nvidia stuff i just think the use case for that is not as big as the current usage of it because of things like the TPU that Google has produced, which is totally fine for most AI tasks.

12:56This is very interesting. I'll tell you what the problem is with this. So I don't think this is going to send NVIDIA broke. I just want to be clear about that. I think NVIDIA is such a successful company generating so much free cash with so much cash that I don't think they're going to send themselves broke. So I don't want to kind of ring alarm bells on that, but I do want to say this a thing called correlated risk. Now, what this deal does is Invitra is basically saying, you can go and buy these chips and a third party bank is going to finance them, but we will underwrite that finance so that there's no chance of them going to zero.

13:37The furthest they can fall is to 25 % of their original value. And remember that 25 % is over time. They're depreciating anyway. It's not like they're going to fall to that overnight. And the problem when you think about it is when would that be the case? Under what circumstances would NVIDIA have to make good on effectively paying money to the banks to backstop a minimum price? And the answer is it could only happen at a time when NVIDIA was suffering the consequences of whatever caused that problem to begin with. So in the most tricky circumstance economically that NVIDIA would be facing, they would have to absorb this make good expense on these chips and this money that they owe to the bank.

14:25That's a correlated risk. They can't really happen independently. Because if they happen independently, if they just go and sell all of these chips to some data center, and the data center goes broke, the price that they sell those chips for is going to be higher than the NVIDIA floor anyway if everything's going well. So it's no problem for NVIDIA. So I think my biggest issue here is the correlated risk of it. But I said to you, I don't think it's big enough to send NVIDIA broke. They're too big. I think this is what's going to happen. NVIDIA's gross margins are going to fall. They're not sustainable over the long term.

14:59They're going to face lots more competition. And fundamentally, the biggest risk continues to be the same risk, which is all of this money that has been invested in AI, let's say that in the broadest sense, can't deliver a return on invested capital that is adequate for shareholders. It just doesn't produce enough of a return. And that imperils the whole stack. And recently I've been trying to listen to a lot of podcasts by very smart and rich technical people or tech investors that you could broadly summarize the podcast with the title why we're not in an AI bubble. And I listen to these podcasts and I have to say, every podcast I listen to just convinces me more and more that we're in an AI bubble.

15:43Like the smugness of the way they speak about this stuff and the flimsiness of the arguments that they're making is so ridiculous. Like one argument was it's not like 1999, 2000, because in 2000 there was all this dark fibre, which means fiber that wasn't being used, but built but not used. And there is no dark data centers or dark chips, which is true, but that's because we haven't built that many data centers and chips yet. But if you look at how many we're planning on building, and so I think the main frame of reference that shifted for me, listening to all this stuff, is maybe the crash is further away than I suspected it might be.

16:24But then I do keep coming back to the same point, which is the trigger for a crash in a bubble is completely unpredictable. and like it could have been um like what meta did like you know meta's vague stuff i actually think meta maybe is undervalued company at the moment but like so they're vague stuff and it could be google turning around and saying we're just not getting the roi on investing all of this money so we're going to pull back on our investment who knows what it's going to be um but i do think that this is nothing special this was inevitable turning this into some kind of asset-based finance or an asset class that was going to be$500 billion or a trillion dollars.

17:03I just felt like this was inevitable. And so, yeah, I just think this is another symptom of the bubble that we're in. I think the biggest, the most concerning part was the point that Ben Thompson made, which I hadn't thought about, was the reducing relevance on CUDA. So how NVIDIA became so powerful is it created this language if you want to program these chips and you had to use NVIDIA chips because if you wanted to use CUDA, everybody's using CUDA. and that's their real competitive advantage. It's their cornered resource. It's super powerful. It's their switching costs. If you take away CUDA, and it is just another chipmaker now competing against all these other chips, like all the hyperscalers are now building their own chips that we talked about.

17:40They've got their TPUs and all the other stuff. So Amazon's building chips and Google's building chips and I'm not sure about that. I think Microsoft's even doing chips. But you shouldn't represent those as being chips of the sophistication of Blackwell. I would say, which is NVIDIA. Yeah, NVIDIA. No one's saying that, but no one needs these Blackwell chips necessarily. You can get away with doing specific - Well, you need some of them, right, to train the model. We're talking about inference here. And as we know, everything's switching to inference. So how do we run inference more cheaply? And then you've got obviously the open-weight cheaper model.

18:12So it feels like this is the Cooter and NVIDIA is the Hermes, but not many people can afford Hermes bags. It's becoming more and more Hermes-like and less and less Kate Spade. And this is just a smaller market for what is this sort of super premium called the Anthropic NVIDIA duopoly is now becoming far less relevant, I think. Well, it's worse than the Hermes example because it would be trying to be Hermes in a world where everyone said, the only reason I buy such an expensive bag is because it's such a good bag. But that's not how Hermes sells their bags. They sell it with the logo on the side because it's sending a message, which is, I think this makes people think I'm rich, basically.

18:57And so NVIDIA doesn't have the luxury of that part of – they're not a luxury brand. They just fundamentally create better tech that was really, really desired and required. And if it's not required, then they cannot possibly maintain these margins. And the thing is this. What's their gross margin? 70-something percent? It's very high, right? And so if people are coming in and undercutting them, like this is a crazily profitable business, they've got a long way that they can drop their margin to keep selling. And I think that that is the long-term trajectory of NVIDIA. Well, no one's suggesting that NVIDIA is going bust.

19:34Even the biggest NVIDIA bears saying they're going bust. They've had multiple 80 % drawdowns in the history, though, when they were a games business, basically, in fact, a games chip maker. So an 80 % drawdown is hardly out of the question for NVIDIA. and it feels like all this circular financing stuff gives you a really bad feeling. Like you shouldn't need to – it's like me giving loans to people buying holidays and I'm giving more and more loans to less and less reputable buyers. Eventually, you kind of run out of money to give these loans to these holiday buyers and the business goes – and NVIDIA is a super strong business with a great moat but it feels like that moat has just become a lot shallower.

20:10Well, I'll tell you the worst argument that I heard in all of these, why we're not in a bubble. It's a consistent argument across all the podcasts. The counterparty to all of these deals are the most successful companies in the history. The hyperscaler bullshit. Yeah. So it might be true. They might be the most successful companies in history. But the thing is this. In order for this to continue, they have to keep spending tons of money. And if they really are the greatest and smartest companies in history, at some point they may well say, we're not getting a return on this money and our shareholders are sick of it and they want us to start making cash again and so it all comes to a halt and this could all grind to a halt very suddenly and i think that that is the risk the two two really obvious risks do you see is one is what you know what causes crash as you said before it's not necessarily things going to zero it's growth slowing so and inevitably the growth slows as you said hyperscaler spends the obvious sort of weak point there.

21:08So suddenly Google realized, and AWS realized, I can't keep spending way more than my cash flow and taking out debt. So that's one, obviously. And that doesn't mean they'd have to go to zero. They can go to$100 billion and spend. Suddenly it spends off 50 % because it's $200 billion now per hyperscala. That's one risk. I think the other risk is the two reasons for this boom is really OpenAI and Anthropic. And these are heavily loss-making businesses. And eventually, if they can't get a capital raise away somehow, or NVIDIA has to come and save them or something like that, then that's the other potential precipitator of the whole bubble just bursting.

21:43And there's two. They seem relatively likely. I think both are likely to happen at some point. The question is when, not if. So you've got this whole massive bubble based on two inevitable problems. If you look at, you know, I always pick on data center operators, but if you look at these data center operators, a lot of them, by the way, there were core wave ads all over San Francisco. no iron ads i didn't see any iron ads so the thing is that if you look at these players you have not entirely but you have a mixture in there of really crappy bitcoin miners that struck it rich it actually reminds me i don't know if you know this story but like these oil magnates when they were um mining or drilling for oil originally like before there were cars and the combustion engine, the oil was just being used, I think, as kerosene, right?

22:41Yeah. It's just a very, like, better than whale fat kind of - It was mostly for lamps, mostly, for life. Yeah. And so they were, like, wealthy for doing that, but they weren't crazily rich. And then all of a sudden, someone invents the combustion engine, and it's like, oh, this oil is unbelievably valuable. And so the Bitcoin miners are, like, a worse version of that. like the bitcoin mining business that wasn't even a business like i don't even believe in bitcoin so i believe even less in a mining business of bitcoin and so all of a sudden they just strike it lucky unlike now we've got this asset that everyone needs and so you've got some of them and then you've got these other businesses that have been around for two years and are worth like five billion or fifty billion dollars and that has to be opportunistic i'm not talking about the the foundation models here like i mean they got their own problems but like you know sharing ai it's like two years old and so like it's two years old by a guy who's like a world poker champion and so that's all right that's a legitimate way to make money but like he knows how to uh he knows about human psychology let's say that and so my issue with this is just every bit about it this doesn't feel like Google 2003, which felt like really smart people building a product that it was obvious why that was such a good product and how they were going to be able to make money from it.

24:09Google didn't have cogs bigger than their revenue. They didn't have these massive token costs that were bigger than the revenue they bring in. Yeah, they barely had cogs. That's the point. That's the best business ever, right? so i'm just i'm just every bit about what's going on now makes me feel uneasy every bit about it except to say these models certainly feel like they're revolutionizing work yeah we'll be back in just a moment with our last story

24:44and we're back and as foreshadowed on this very pod canva's institutional backers have quickly reduced its valuation mark on the australian decacorn with friends in the pod blackbird and airtree wiping a whopping 10 billion from their internal bowels uh the afr's tech section reported last week that the Blackbird and Airtree cut their value to$34.9 billion after commissioning external assessments. Canva's internal valuation, which is a number it uses for employee share issuance, has fallen following an independent third-party valuation from$39 to$31 billion. On public markets, interestingly, Adobe has fallen 23%, Figma is down 68%, and HubSpot is down 73 % over the past year.

25:24HubSpot is an especially pertinent comp with the SaaS business reporting US$3.1 billion revenue last year at a 20 % growth rate. These metrics are almost identical to Canberra. HubSpot is likely more profitable than Canberra as well, generating US$45 million last year. HubSpot valued at only US$11 billion. So that is a pretty scary comp for Canberra investors. Adia, do you think that's a fair comp? Well, firstly, I want to correct you and say we didn't foreshadow it on this podcast. We predicted it on this podcast. That's the first thing I want to say. The second thing is I want to disagree with you about your pessimism in one respect.

26:04The thing about public markets is, you know, you love the Benjamin Graham quote, in the short term, it's a voting machine, in the long term, it's a weighing machine. And when it goes through the voting part of the voting machine, like Saspocalypse would be one example, then it becomes an inefficient market. Like I'm not a believer in an efficient market theory at all. I think it's a load of rubbish. And so it becomes an inefficient market. And I think public markets at the moment are inefficient at valuing these businesses. And I want to explain why. It's not just an emotional reason. It's because if you look at the revenue, and like Figma is the greatest example of this, if you look at the revenue of these SaaS businesses that are considered to be most imperiled, actually, they're still doing very well, by and large.