Hey everyone welcome back to another episode of the meet kevin show. Today we have the honor and privilege of talking to the director of research brett winton over at arc. Invest. I'm super excited we're going to talk about a whole range of topics from, of course, the fed inflation wage price spirals.

Of course, we'll talk about zoom and maybe we'll touch on palantir and roku and, of course, the future of innovation, hydrogen electric vehicles. What's going on in russia, you name it, we've got a lot to talk about, we've got an hour, so, let's get into it. Brett welcome aboard the first question i have for you has to do with the fed's big u-turn uh in december, which we learned about in their minutes on january 5th. They came out, and you turned entirely telling us that inflation was no longer transitory, that we were heading into a disaster of broad-based inflation, more persistent inflation, suggesting that it might be necessary to have a faster runoff and a faster tightening cycle.

What did you all make of this said arc, especially since right after that we saw a continuation of valuation compression? I mean broadly. I think that it's evident to us that over reasonable time horizons technology is, is massively deflationary and that over time, horizons that are uh. You know consistent with our investment time horizon that that that's going to manifest in basic, cheaper products for consumers, uh and so the um. I think, there's a lot of commentary out there that thinks that the fed is behind the curve they look at.

Oh, the fed always has to like over correct in order to stave off inflation, but if you look at the sources of that inflation, particularly early on a lot of it, was concentrated in the automotive sector uh. You know in both used and new car pricing and we think that robo taxis are going to price at equilibrium at 25 cents per mile and so right now, when you buy a new car you're, paying 70 cents per mile for your transport. And so that's going to wreak all kinds of havoc on um kind of both the new car pricing and used car pricing, like what good is an internal combustion engine used when, when you can literally travel cheaper by getting an uber-like experience without a driver that tries To chatter at you, and so that's like one example, but across all the technology areas that we focus on over the time horizon that those technologies are going to infiltrate the market. Consumer goods are just going to get much cheaper uh, and so, if anything, there's like there is, there are some downside risks to deflation.

But i think that's over the course of the decade going to be more of what people are worried about and thinking about, rather than uh kind of, like the supply chain disruptions that the fed is responding to in part, because they politically have to respond to it. To show that they're doing something and then given what's happened in in uh in ukraine uh now you know after everybody was falling all over themselves to say the fed's gon na tighten eight or nine times this year and now the fed's likely to back off, and So um and and then and then from from uh, like if you think about the game that we play, we try to play the weighing machine game, meaning like people say uh in the short-term markets, voting machines in the long term, they're weighing machines. Well, we try to underwrite positions and technologies over a time horizon that we think is consistent with an equity holding, which is five years plus uh, and the the the market has been really playing the voting machine game over the past six months. If you look at correlation of um kind of innovation portfolios to where, on the yield curve, they're correlated they've actually gotten much short term more short-term correlated as in they're, responding to that incremental fed move and they've lost their connection to the long end of the curve.

Um but the weighing machine like the discounted cash flow uh that you're getting should be discounted by the long rate, not the short rate, and so, and so it's it's clear. That kind of, like the the marginal pricing, is being driven by flows, not actual fundamental analysis, which is why we think innovation right now is on sale that if you do good fundamental analysis of intangible assets over reasonable time verizon, you can identify massive mispricings and then Take advantage of this so in some ways it's a really great setup. As far as we're concerned, uh and um, you know the everybody that's been long. Innovation has gotten squished over the last.

You know three months uh and so that's those are the puts and takes so you said, valuing the intangibles. What's your process, for that i mean like, for example, gm bought, a chevy cruze right, they've got the whole self-driving. Suite itself is probably worth what 40 or 50 of what gm is worth. Meanwhile, you got the ceo of gm and ford according to a bloomberg article here recently, supposedly tightening supply chains on purpose to keep their margins high to squeeze more profit.

How how do you parse these things out? Well, i mean, i think, there's we have a framework we go through where we always start at the top down. We do careful work on on cost declines in technology and then unit economic cases of different buyers in different sectors as costs come down and then at the individual company level we go through and assess people, management and culture and product leadership. The barriers to entry, where the company sits in the value chain, their ability to execute uh and oftentimes uh mature incumbent providers will talk a good game and, and they may even make acquisitions proves cruz - is an amazing technology? Don't get me wrong? I think there's questions about its scalability and their ability to manufacture scale, the robo taxis they need to to extend that technology into commercial operation in real in a real breathy way. Um and um marginally.

As i understand it, what's happened inside gm is that the ceo of cruz got forced out because the gm mothership wanted more control over that asset and, historically, an incumbent um, a a company with uh uh incumbent's dna uh when they take more control over uh. In an innovative arm of the company it does not lead to good things now. Kyle voigt is is back and back in charge. It was just announced today he's ceo of cruz now uh and i think he's really talented, and so i you know it's possible.

They'll be able to uh continue to aggressively pursue robo taxi, even within a more firm corporate embrace from gm um. But the the history would suggest that the company with old dna is not going to successfully make the transition to uh a new innovation plan, which would be very rare, is that true of like uh like an intel as well, you know now they're trying to move The entire industry, the entire company, from from a hardware focus to a software focus and the last earnings call they didn't want to talk about q4 earnings at all. All they wanted to talk about was their wonderful service transition. Do you have that kind of same feeling? Yeah i mean intel's interesting in that they i mean for one thing: they they they miss the transition to mobile and and you'd argue they've missed the transition to ai as well um and so the the um.

You know the the company was on a real, sustaining innovation path where it's like. We can continue to try to pursue moore's law scaling to um, to deliver better computation per unit price to end users, uh and um. The the miss on the transition to mobile was realizing that that actually, the um, more compute performance without paying attention to the power efficiency of that compute performance would miss like, would not set you up to dominate a mobile platform uh and then on the ai side. Uh, you know the the parallel processing requirements for doing massive ai.

They just haven't set themselves up, uh and and so um often with technology transitions. The the the vector that um a technology is competing on, runs orthogonal to the existing vector, and so the incumbent um just doesn't uh can't can't wrap its head around how it needs to compete and, and often it means they're mis-staffed. I mean back to gm and ford, you know they. They now have finally recognized hey.

We need to invest literally billions of dollars in ev. This is not just a niche category. This is going to become the category uh, but they don't have the human capital to to pivot. To doing so, you know you need software engineers, not mechanical engineers.

You need mechanical engineers, but the marginal important person is the software engineer right and so, if you're dealing with software-defined vehicles - and you have a and by the way, i'm a mechanical engineer by education. So i i love mechanical engineers, but, like you, need to have a workforce that is able to aggressively pursue um the technology that is going to be the differentiating factor for your product or service, and they just don't have it uh, and so it's it. You know can, can you then effectively spend that billions of dollars that you realize you need to spend? Probably not it's gon na be very inefficient. You're gon na hire lots of consultants, you're gon na bring in bcg and mckinsey and they're gon na say what do we do and so then it's like you're setting cash on fire trying to make this transition and then in the meantime, you have a big core.

Like very expensive business running that you can't turn off because you're still trying to pay dividends out of the shareholders and - and so it's really messy and complex and and hard to execute on, and so could they do it it's possible. Would i bet on them doing it yeah not really possible versus probable? Of course now it's is this. Why you think that i'm, assuming you think this, i don't want to put rewards in your mouth, that you think gm's less likely to scale than tesla, maybe because they have those software engineers that vision-based guidance versus the map-based guidance. Or is this sort of the similar wavelength of thought? Well, okay! So there's two: do you mean scale robo-taxi or scale um the ev business? Let's do both, but let's start with robo taxis, because that's what i meant yeah so on robo taxi.

I think there's a there is a very um real and pressing question um. That is uh. If, if you think about cruz's strategy, is we have a big uh hd map of the world and we have very expensive sensors on our vehicle that we can have the vehicle be navigating through the world and and match itself to where it is on that map? In a really precise way, and so anything that deviates from what you see or what you sense on distance with their lidar and the map of the world, you know is something that's dynamic, it's something you need to pay attention to, and so that actually seems like. Philosophically a good way to go about solving the problem is, is you collect like a big, basically virtual map, and then you know exactly where your car is on it? And so then you know where you can drive and where you, the problem is that it doesn't scale very well in that um.

For one thing, any given intersection on that map will change, because there's road construction there's all kinds of things that happen and probably change over the whole breadth of the map more frequently than than you realize, and so then, when your system encounters those changes, then the System degrade the system, performance, degrades and those those change circumstances are probably the hardest areas to drive. So a road construction site is harder than a normal road to drive. So your system degrades in the exact spot, where you need your system to be most performative, and so, if you're, if you're designing to like a performance condition where, like you, can't fall below some threshold, which is human driven performance, um, then you're better off designing yourself. Um into the weak environment, where you, where you don't know anything or you know very little about the existing environment and then having uniform performance across the environment, and so then that's tesla's philosophy is.

We should be able to drop our vehicle into any spot and it should be able to just based on visual information figure out. You know where is drivable and where it's not and and how to follow the rules, road, uh and so the problem with that is. It's a really complicated and hard problem and it's a problem that you need massive amounts of data to solve uh and it's a problem that it might look like you're, not making very much progress but like the rocket ship. You know you you're, hopefully, on this exponential slope, going upward as opposed to a slope.

That's asymptoting to some performance consideration so right - and it's also it's also tesla. It's not that tesla doesn't use maps, and it may be that they need to use higher definition maps than they're using because they're routed. It's not like they're guessing where to drive. It's not like.

They are like wandering around like a maze right. It's a question of how um, how much information density do you need in those maps in order to perform and the the so then, in terms of scaling, not only are you, you potentially hit a performance constraint in any individual city if you're using the hd mapping Approach, it also means it's very hard to go from one city to another city if you're using that hd mapping approach, because then you have to map that entire other city - and i know some implementations would literally mapping - is not just hd mapping. It's like going in and manually telling the car where it needs to stop at a stop sign in order to get good sight lines to know if it can turn left, and so who does that? Not i'm not sure i should disclose, but it's a startup that i respect and and does really well i mean or had good performative technology yeah, but but i had the short button ready. You know yeah, no, but but think this thing like so it's it's not.

I'm not even sure it's main it may be that they are operating then, with more than just hd maps. It's hd maps that are heavily annotated right in human annotated in some way, and - and you can imagine that's how to get to market - that's what you might need to do, or what cruise cruz has decided you need to do is this. Is there? Is this intersection that our cars always have a problem with? Well, how are we going to solve that? Well, you can try to figure out how to solve it in a generalizable sense, but maybe it seems impossible to you so then you say: listen, let's! Just like manually say, the car should not proceed beyond this point without stopping and looking, and that would be a solution to the problem and actually that's a great solution. So you do it and it works, and you say: okay, we're just gon na do that.

But then that intersection changes slightly and it's like oh they're, failing in this intersection again, you have to change the stop point here and soon you have to hire thousands of people to be like trying to like manually tweak the things, so it still works throughout the Whole system, and so it just the the so your fail modes grow exponentially as you try to scale to different geographies and so figuring out a way to to design yourself out of that. Inefficient scaling is really challenging and it could require you to not like to solve the problem in a narrow sense, because you are, you are concentrating on the critical path to generally solving the problem. So so i think that's you know yeah, and that was that was really insightful. Thank you for that.

So why, though, if you're gon na drop the tesla blind into an area and teach it to respond, why not equip it with a forward-facing lidar, just give it a little bit better edge? I mean hey yeah humans. Look at night, i wouldn't mind having a lidar. Some night vision, i mean why not it'll make a little easy for me. Well i mean it comes down to like.

Would you rather have uh? You know tesla selling a million cars a year or like fifty thousand right i mean so it's the cost of dollars. Yeah, well, i think that there's a um you want to get. You want to place a bet on the sensor suite you need to successfully commercially operate and not exceed that sensor suite because of the the cost it will impose on the consumers of your vehicles. In the in the subsequent limiting of your addressable market uh and so like, if anything um, i would expect well who knows given their recent commentary.

But it's likely that once robo taxi is really working, the the unit size and the upfront cost of vehicles continues to decline. Much lower than people really understand, because you would rather have imagine if i'm calling a robo taxi and the robo taxi system knows. If i'm trying to travel to like san francisco or just like two miles away on city streets in l.a, well then it'll dispatch to me the vehicle that's well suited to that. To that drive right, and so i don't need to build a vehicle that is highway.

Spec to to trundle people around for two mile trips, and in fact it's probably going to be optimum for me - to produce like little people, pods um, that are neighborhood electric vehicles, because those will be lower cost to manufacture and then i'll have an environment. That's much more sensor penetrated and so i'll have a higher volume of data feeds to continuously make my system better uh, and so the there is a real. You need to have robust, uh kind of inexpensive assets to maximize the data capture or the data. Optionality that your system has and that's what they've they've weighed so um, you know if they if they saw an advantage on an incremental basis of adding liars to the system.

I think they do so. I think it's i i agree with like elon's take, which is basically like. The road system is designed for vision sensor, like that's all of the every every single design consideration made in the road system is. Can this be operated by a thing that can see, and so, if you have so then, like vision, sensors should be um.

You know necessary and sufficient to deliver autonomous driving and if you rely, if you have to rely on some kind of like laser range finder, that gives you the precision distance of the vehicle 300 yards out like then then you're actually over engineering. Your solution to the problem, uh and and subsequently restricting kind of your your actual path along. You know critical path solution. That's that's very insightful! Thank you for that now uh and i love the line necessary and sufficient.

I think that's! That's so important because, like you said, it's got to be affordable, otherwise you're limiting your data anyway and, quite frankly, tesla's autopilots, as we've seen through the data safer than regular drivers anyway. Already right, so i want to move to battery scale. So here's the thing uh. We 90, i think 90 plus percent of uh lithium iron phosphate batteries are manufactured by china, uh the lithium, cobalt ones.

We've got much more dispersion here. Problem now with russia, and ukraine is some of those supply chains are coming under question for lithium and cobalt. Sorry, not lithium called cobalt and nickel and uh we're we're this there's this consideration of. If tesla goes for these lithium iron phosphate batteries, maybe other manufacturers will follow we'll go for these.

Maybe cheaper batteries maybe a little bit better thermal resistance, but are we now giving too much power to china sure it's great for tesla, because they're close to catl at with their shanghai factory? But you know: when do we adapt uh in the united states? What companies are going to catch up? What about? What are we going to do in germany? Are we focusing on the 4680s, or do we just go with cheap about the cheap batteries for now? Well, i think tesla said that you know their standard range vehicles going forward, we'll have iron batteries and uh. I think i don't know if tesla would agree with this, but but it seems likely to me that, at least in the immediate term, marginal production of evs is going to grow through the lithium-ion phosphate stack uh. If you look at so we've done, caustic lines on both cobalt batteries and lithium-ion, uh, phosphate batteries and lithium-ion. Prostate is actually a less mature technology in terms of uh production.

Doubling so um kind of you you for a marginal increase in lithium iron phosphate batteries. You actually get a higher percent cost decline than you do on the cobalt battery side. That's the problem because it's less mature now yeah, because it's less because the production production scaling hasn't happened there as to the same extent as as cobalt batteries, um, okay. So the problem for other manufacturers x tesla, is that their um drive train systems aren't efficient enough to deliver a you know: happy mileage range on those iron phosphate batteries, so so tesla has this vector by which they can continue to expand um production.

That's for now is closed off to other manufacturers is as well wow and um now, insofar as i think any um, you don't want to be a single source to anybody or single geography in any kind of supply chain. Luckily, like lithium iron phosphate, it's not like the raw materials are scarce, so you know you can go to australia for iron. You can go lots of different places and i would imagine that, as just like with the 4680 cells, as it's demonstrated, that this is a viable mechanism for getting ev supply on the road that other battery manufacturers will step step in and begin to produce iron batteries. Uh and so um it it's yeah, but but i do think it's like it's an interesting wrinkle right now, because you know i my sense is the other manufacturers are they're like okay, we're going to invest.

You know 10 billion dollars in ebs uh. We need to lock up batteries so then they go and try to lock up like cobalt supply out five years at a fixed price and uh and - and that seems like a way for them to de-risk the the supply ramp right. The problem is that then they're locking in a price on a battery stack that puts them in a much more expensive cost of goods cost of the vehicle relative to someone who's being more agile and can switch to another battery chemistry, uh, and so they may end Up four years out, you know they're great, you know big push into ev plans playing out and they're stuck at this massively higher cost structure, so they either are um. You know profoundly unsuccessful financially for having done so um and or uh tesla is massively profitable because, there's, like you, know this huge price or cost umbrella that they're sitting under uh.

So so it it's unclear to me that this how the setup works out if anything, so we have an expectation that 40 million evs are going to be sold by 2026. um and the risk for that is definitely the exclusive tesla tesla traditional oem supply, like they. You know now, since, since we like uh, you know, we said the same thing last year was like the risk is the traditional oems and whether or not they're going to be able to ram from here to there. They have all announced massive capex plans, like they've.

All basically like pulled forward their targets, they're we're going to dump a ton of money on this and it's still the risk. I think, because that execution is it's it's it's really clear to me that they didn't. They didn't actually believe it until at least you know 24 months ago, and and that's it's not clear to me that they have the wherewithal to actually scale it in the way they want to now. There are a lot of folks who say they would prefer to bet on commodities just because they look at the last.

You know six-month charts and we see uh. You know nickel aluminum all these commodities taking off. Is that a big mistake when, as you mentioned, a company like tesla could just be agile, why do i have to buy from you or even that material when i could use something else? Would that be a risk for betting on ev? So we look when we do our forecast. We look all across the value chain for what we think are good opportunities so and and in part, to de-risk the forecast.

So you know part of doing our ev forecasting is to look at okay. Let's look at the lithium, you know and the cobalt and are there efficient investments there? Is there uh? You know what is the spot pricing look like? What is the cash cost to produce these materials uh and with lithium we did have uh investments in lithium companies? For a period of time, and then the spot price got way out of whack to the actual cash cost, to bring on more production and uh, and so we took profits into a part of the value chain that we thought was going to um. You know approve significantly more cash over time, uh and so like and within the commodity companies generally. So there's a difference like trying to if you're gon na go out, i don't know, buy yourself some raw lithium and keep it on your shelf.

Good luck to you! You know there's like some inventory problems there, but, but you know my experience with um commodity producers is that usually the executive teams make very poor capital allocation decisions right at the wrong point in the cycle. As in it's kind of like this, this narrative infiltrates the market where oh, the world is short lithium, and so then suddenly they, you know, try to roll up the whole industry or monopolize the market in some absurd way that ends up incinerating a lot of shareholder Capital, that's not uniformly true, but it's you know it happens. More often than you'd like it to uh and generally like our style is, is to um look for kind of the the the real innovation enabling portions of the value chain and, like you said it's like you, could have two years ago, looked at the windscreen, like, Oh cobalt's, really in short supply, i'm gon na go and like buy a congolese cobalt miner, which there are a few and there you're taking on a lot of human rights issues. To be frank, and it's like a uh, you know there's a lot of de-risking.

You have to do to be comfortable putting money into that sort of investment and then meanwhile, like i, i think, the suddenly the marginal demand for batteries is probably not going to come through football. You know, and and tesla recognizes it and, like you know, is wildly innovative and is is identifying all the choke points and its ability to scale and figuring out routes around them. And so, if your bet is like, oh there's, this choke point that they can't get around, and so then that's where we're gon na get like this massive return, you're you're at great risk of there being like a you, know, shortcut around your spot. I mean it's just basically like battery day battery day, like you mentioned cobalt shortage and battery day, what do they do? New battery a fraction of the cobalt needed right exactly and - and i mean i think, that the um yeah it and i think you also have to be agile, like or or part of, the reason you know we look at the company's management and their in their Product leadership is also trying to measure um, whether or not a company will be able to route around kind of the volatility that comes with uh uh, the innovation, strategic landscape, things are so fluid, you know, and and um as you're and there's a lot of kind Of like ambiguity and uncertainty as to what the right pathway will be uh at any given point, uh and and and so um yeah, i think, generally like in in some ways the simplest way to put it betting on like a raw material price inflation is, is In some ways, betting, betting against innovation and that's the opposite of what we try to do.

That's a good point. Well speaking about inflation uh, you know i've been reading a lot of earnings calls here and it seems like almost anything. That's customer facing like kimberly clark, kraft foods insurance companies like travelers, progressive under armour wing, stop advanced, auto parts home depot, the banks, labor inflation, they're, all complaining about freight labor costs going up, uh and and then of course, part inflation. Uh input cost inflation.

It is it possible that if we keep seeing this that we end up getting some form of wage price spiral that even when our supply chains improve, we end up having inflation stay more persistent and then the fed needs to become substantially more aggressive. And we end up with sort of a mini, paul, volcker yeah i mean, i guess if you look at the underlying data, like consumers are coming back into the workforce. So there's not enough. I don't think there's enough tightness in labor markets right now to actually allow that to occur.

There's a lot of people that kind of stepped away, probably on the basis that they've made all their life all income in dogecoin or something that thought they were good. And then it is no longer worth nearly so much and now the wages being offered are like a little more attractive, so they're coming back into the world um. So you actually think this compression in in stocks and maybe some some alt investments could actually motivate people to come back into the workforce yeah. I think so i mean, if you think about, i think it was the bank of america call where they talked about how their lower income customers lower income, or - i guess people that had two thousand dollars in their account now have ten thousand dollars.

You know, i can't remember their exact phrasing, but it was basically like the lower dollar accounts had a lot more in savings than they ever have, and so i think, generally, you know. Uh consumers got their stimulus checks, they stopped spending on services during the pandemic, and so then they had balance as a crew and some of them you know, invested it and and did a lot better than they probably thought they were going to and suddenly felt very Rich and then underperformed and now they're like oh, i i guess that's not as much money as i thought it was uh and it turns out. I can't live off of that and now i'm out going out to restaurants again, which i wasn't before so maybe i need to get a job and i think that that's likely the way it will play out. I think, over the medium term.

You know the way we think about um, ai and automation, and how it's going to play into labor markets is it's basically hey. I should be thought of as augmented, intelligence, not artificial intelligence and and what it means is that it'll be for a knowledge worker you'll. Be basically amplify the power of that workforce and productivity going up, yeah, yeah, yeah, massive productivity gains and, and so the um you know, the the marginal value of an employee might actually go up to an organization because but but the productivity they're delivering will go up. Even more, and so the net result is is, is actually much higher gdp growth than people anticipate, uh and so on.

Uh um, you know so so it it like implies like. If you deliver lots of gdp, then you can't end up with that. Um wage price spiral because you're you're delivering much more in product to consumers, wow wow, that's incredible! So how long, though, until we we see uh these sorts of improvements, uh in potentially productivity, but you can see well. I think that there are issues with how productivity is measured in the gdp accounts but like if you look at what openai is really saying and the um the alpha code from from um deep mind um.

These are just from a software engineering perspective right. The the alpha code example, this is a program, an ai system that can take like written um, um kind of programming problems and perform better than the median competitor in uh coding, competition that is literally used to give people jobs right and so, and it's not like Performance so so i talked to software engineers and they're, like we saw that we're like yeah. Well, you know, and so what it means is actually like. The the role of the software engineer needs to be like kind of higher level.

It's like more of a race level yeah well, and it's like that software engineer can can marshal. The resources can can turn these coding ais towards the problems that are most important for the organization to solve and vet them and audit them and figure out how they puzzle piece together. So it's not that the java software engineer will disappear, but the volume of effective code that's being produced will increase right and, and so the the degree to which you know think about the automation tools that we have right. Now.

You probably use some automation tooling, to run your global enterprise over there right. It's pretty clunky hard to stitch together. It frequently breaks like you know where it's it's really rudimentary. If you think about it, and so like you as an as an enterprise of one or you know whatever like, you will become much more powerful like much more efficient at like just getting to the interview, then then all of the stuff you have to do on The back end to promote yourself and everything else right and so like that will apply to every knowledge worker will roll across the economy.

And so you know if, if meet kevin, is putting out uh, you know 20 youtube videos a week as opposed to just seven. That's a massive productivity improvement. Maybe that's what you can do no, or you probably do more than seven weeks. I don't know how many well.

Thank you for that yeah! That's! So if it isn't, i mean i suppose i mean this is incredible. I suppose it is possible, though, that companies might just say geez. We used to need 50 engineers. Now we just need 20..

Isn't that a potential risk for for some engineers, as well just thinking separately from inflation, now, of course, yeah uh, the the history would suggest. No, as in okay um, you know like so we looked at, for instance, um after the we we looked at like automation and its impact on jobs. For one thing, it's always always always every technological like cycle and innovation. People basically claim that it's going to cause technological import on unemployment so like.

If you look at the agriculture sector, you would think well introducing the tractor gosh that you know you need much fewer labor hours to to work a field so clearly that causes unemployment. For for all those agricultural workers, but instead what actually happened is one you uh. They shifted to other higher value, jobs, uh and then two, the marginal labor hour being shrunk on the in the agriculture workspace, was not actually the hired hands. It was the owners of the farms who were doing essentially unpaid labor to make them happen.

They were relying on you know, they're having five kids so that they had enough free farm hands to work. The fields well like the number of hours that those five kids had to work reduce because the tractors could do that work, but they still hired the same number of people, uh and and so but then there was massive economic production and so uh. You know people moved into industry and everywhere else, uh and sure you grow capacity and everything yeah and and and so the you know we take take back to robotech right. You would think robo taxi gosh.

This is self-driving cars. That's gon na cause all these drivers to go out of business like somebody who's a professional driver. What job are you going to do once robo taxi takes over? Well, we actually think the widespread robo taxi system will create more driving jobs than it eliminates because think about who's driving. Today it's 99 amateurs, you and i are amateur drivers, totally unpaid for for doing so right, but the marginal cost for us driving our own car is around 30 cents a mile.

Even if we've fully paid off the car and everything just the cash cost, if you're driving internal combustion is 30 cents a mile, whereas robo taxi, we think at the margin is going to price it 25 cents, so literally you'll be saving yourself money by not driving The car that you have in your garage, uh and so a lot of those amateur driving hours are going to go to something where you're paying a robo taxi service for doing it, and that robo taxi service is going to have to have some back-end set of Operators that are overseeing that system, so maybe it's false to call them drivers, but these are the people that, in the event, the system is a little confused like waymo has them now you press a button, and somebody gets dialed up who like looks at the sensors, Is like oh, it needs to navigate here and like routes it around the problem that it has uh and so the way in which we've done the our estimates, for the costs of that network is to assume that there's some human labor component that has to oversee It that has to be basically like the monitor of that system and because you're shifting so many miles from amateur drivers into actual market behavior uh you create more jobs than even if you eliminate all of the current professional driving jobs in the uh yeah yeah. Just because now monetarily, it makes sense for you and i to just never bother getting behind the wheel again exactly and and that's a lot of innovations actually work that way where they take non-market activity and they turn it into market activity. So the introduction of the washing machine right: what did what did we? What did the washing machine do? Well, a lot of housewives mainly were sitting at the scrub board and washing clothes every day. Well, that was not they weren't paid for that that doesn't show up in gdp statistics right.

But if you convert that into i'm buying a washing machine and then paying a service technician and paying the electricity costs of that washing machine that shows up in gdp and then that housewife, it has free time to join the labor force, which is mostly what they Did uh and so the the um you take like amateur washing and you turn it into professional washing with a machine now if they could just make the stupid clothes folding machine. That would be even better, but that's aside, you know, and and because i'm an amateur's clothes folder and i'm not very good at it. So i i definitely need a professional mechanism, you probably don't like it. No, i don't it's a lot of time, and so it's a same with robo taxis like you have literally an hour a day that americans spend driving you know and and you convert it into and it's not captured like nobody.

You know nobody recognizes that as economic loss, but it's a big economic loss, it's much bigger than than than um kind of the the loss from driving the vehicle like the fuel costs uh and you convert that into robotech and and then it gets recognized in gdp And our view is it's going to be a on on an aggregate basis? Uh, i think it's a 26 trillion dollar gdp creation event over 10 years, so measured against any technological transition in history, at least that mckinsey global institute looked at. This would be the the biggest uh gdp productivity advance attributable to a single innovation bigger than the steam engine, bigger than um kind of the the computers and bigger than kind of automation in the manufacturing sector. Um because we think it's going to concentrate in a pretty short time frame and because and because it creates it's basically like you know, i am not going to - we don't think you're going to pay for the full value of your time for using that robo taxi. Generally, consumers they pay roughly half to get their time back um, but along the way you go from being an amateur driver, who's unpaid to paying somebody else to drive you very inexpensively and efficiently and while you're sitting in the car, you can do other economic activities.

So conceptually i can work on. You know financial models. Realistically i can watch you know narcos on netflix right either of those or gdp creation events either of those um. One accrues to netflix's benefit.

The other accrues to arc's benefit, but the the you know the net result on the back end is that you get the double whammy of gdp creation, because you're paying for this platform that didn't exist before and then you're, freeing up my time to either also consume From another platform netflix or do more work, creating productive output for the overall economy, one more contract written for real estate, one more whatever interesting, so uh. I want to ask you so uh. You know. Some companies here recently have come out with uh guidance, a little bit lower than expected for q1 uh, like etsy 10 lower guide amazon gave a lower guide.

Zoom today gave a little bit of a lower guide. Is this uh? Does a lot of this just have to do with fed uncertainty? Does it have to do with war uncertainty, or are we seeing a slowdown of the consumer what's happening? Well, i think that generally wall street and companies struggle with how to deal with very tough compares. So if you go a year ago, everybody was growing through the pandemic, so zoom etsy, uh amazon. They all had like zooms with something like 340 percent growth that they're growing against and next quarter, something like 100 and something percent growth that they're growing against.

And so you know the fact that the fact that these companies are growing at all after that huge adoption bump uh is a sign that that you know you you would, if, if it like we're clearly back to almost the normal times pandemic, wise yeah, right and So and generally, what happened to our adoption curves? Is it just like pulled forward the model by some amount of time, uh, and so the fact that it didn't pull forward the model with reverberations, but instead they're continuing to grow off the back end of that, i think, is actually wildly bullish for the underlying technologies That are associated with it um i you know i i so i think it's hard it's it's really it's hard to under stress or yeah, it's hard to like the the adoption surge that happened during the pandemic and the fact that we're kind of like okay now We're quote-unquote back, i mean exception of the war going on aside um, yeah and they're still growing. On top of that, i think it's actually a strong sign that the business models that they've set out and and kind of the utility to the end consumer is is very strong. So i don't know if i'd mean that i don't know if i'd read consumer weakness through those particular results. I think that it's that it's certainly possible that um, the consumer as a whole is feeling like marginally.

I mean, i think the survey work suggests they really are feeling marginally more nervous here, and you know you could um end up in in some sort of recession based on either fed over tightening or um. You know what the war is doing to the energy pricing landscape um. You know, i think that there's a lot of kind of macroeconomic reverberations that it's they have to feed through to the consumer. At some point, i just don't know if those are the examples to look for to see them.

What what odd, uh or likelihood of recession in the next 12 months are you all pricing and elon musk seems convinced, there's going to be a recession this year? How did you all respond to that when you saw that? Oh, i didn't see them say that, but i mean i think i mean so maybe we independently came to that idea, but um i don't. I don't know if i don't have a probability on a recession and i like within our like investing framework. It's not you know, since we invest in companies that we're underwriting over the course of more than just a single business cycle, it's not important for our investment process to necessarily know. Generally, we try to like at the portfolio level we, you know during risk-off periods, we tend to concentrate the portfolio and then, as kind of like you end up in respond periods where the underlying securities appreciate substantially, then of broaden the number of holdings in the portfolio.

But that's not informed by necessarily a hey we're going into recession now versus not. I. I do think that there's like strong evidence that the consumer doesn't feel comfortable right now and it would seem reasonable to suggest that then they're gon na pull back on on spending and um. So then, that's gon na take some of the edge off of the um people's ideas about fed tightening but um yeah.

I don't know, but at that same time i suppose, if consumers do pull back, then we i mean it doesn't take much with these high comps. Like you mentioned for us to see, gdp potentially turn negative for two quarters: yeah yeah, i'd say 25. Those are my odds: okay, okay, the um, the quote by the way, just to get your reaction. Um from it was from december 30th.

Uh elon musk predicting macroeconomics, is challenging, to say the least my gut feel is maybe around spring or summer 2022, but not later than 2023. In response to, when do you think the next recession will be yeah? I mean i agree with him, also on macros being very hard to forecast, like i think i think like to me. It seems like the macroeconomic environment relative to what's happening in ukraine, like the dispersion of potential outcomes. Here is very broad, but it's i think we don't really know what the full set of ramifications coming out of that conflict creation are going to be, and the central banks are certainly unprecedented um.

I i just don't you know and like i yeah so so so i i think it's it's it really! It's really it's it's not that it's unfortunate! It's just! I haven't done enough work to really uh forecast. It do you care about the 10-2 yield curve or too manipulated right character. Well, i mean, i think one thing that's interesting is if you look at - and i mentioned it before, but but the innovation portfolios have become much more correlated with the short end than than the long end uh. And if you look at the login, it's actually come down like or you know it's the the ten year is is not nearly as much as the front end of the curve and the curve could invert uh.

But if you're, if you're playing the weighing machine game like the um, the the um, you know you, you should be discounting cash flows at that long end. So i think that you know the. I wouldn't be surprised to see an inversion in the yield curve and that would be a strong sign that we're entering or or potentially tipping into a recession. I think that's like what could happen in some ways.

Kind of what's happened in ukraine, meaning then the fed can be marginally looser. Even though energy prices are up because there's like a pointable reason for energy prices to be up right, uh, you know maybe reduces the risk to some degree, but i it's really hard to call uh yeah. I think that the i do think that um. I think that others have commented on that that we strongly believe in is that all of the supply chain stuff, if you have a massively unprecedented supply chain, um kind of constriction, then you almost necessarily get massively unprecedented, double ordering and call it misbehavior of companies that Are trying to uh de-risk themselves against that supply chain thing and that almost certainly gets massively overdone, and so then, once some of it doesn't even have to be the whole thing unlocks, but certain portions of it can unlock.

And then you end up with an inventory glut on business balance sheets in that particular area, uh and maybe even spilling onto the consumer balance sheet that can result in really choppy kind of performance results for businesses. So i think again it was the bank of america call, maybe the chase call they were saying. Businesses were taking taking out loans, loans specifically to hold higher inventories to protect themselves against the supply chain. Well, it's a borrowing against inventory and running intentionally.

Higher inventory looks fine as long as prices are going up, but when suddenly it's like everything's on sale all at once, then you have big inventory write-downs that aren't great for your financials. So i think that's that's somewhere to come as well. Yeah! That's that's! That's very interesting uh what um yeah, because even recently, some folks are saying that oh well, maybe those inventories are in transit, but okay, fine they'll show up eventually, even if it's taking longer they're still coming so uh, okay, uh! So some so maybe uh rapid fire. Uh, individual tickers, here, okay, you know this one's coming palantir.

What happened sure i mean like i described as we're in risk-off periods, we tend to consolidate the portfolios and so um. You know we had some concerns about their competitive positioning within the government space and we had better risk money, elsewhere, uh, and so it's not. You know they're when you're making a decision to like consolidate a portfolio. It's not like those are easy decisions, but you look at kind of like the risk return profile for all of your positions and say: what's lowest on the stack rank relative to something else that we think has comparable upside.

That makes sense so so in risk off periods to make the simple. Basically when prices are going down narrow into the ones. Maybe you have the most conviction in. Is that that's what you're saying exactly yeah yeah? If we have a better edge somewhere else, then then we'll operate somewhere else? What what did you think about their idea of investing into spax to uh the palantir to to basically try to vc partner with them and then get that software contract is this? Was this a little bit of a gimmick? Was that strategic, you know or is the bigger focus on government and maybe their government activity not growing, because that is something that stayed stagnant yeah.

I think for us, it's more um kind of. I think that they had a really differentiated product in government. A few years ago and then the their competitive positioning has diminished and then the government has those government contracts. I think that they've shown a higher interest in multi-sourcing.

So so you know it's an interesting company. I think that the you know the technology is super interesting. We're very bullish on ai generally, i think that the um marginal competitive weakness on the government side just left us with you know other places to go. That's fine! All right! Let's move on from that one! Let's go to uh roku, so i read the roku earnings.

Call in the roku earnings call, they talked about how competitive, uh, manufacturing tvs is and how complicated and frustrating it is for them to get their hands on chips. They got to spend more man, hours or people hours. I should say on on chips to re-engineer them. So he could, they can use different chips in their tvs.

Complaining about you know all the supply shortages, all the usual uh. They also said how tvs are a really incredibly low margin, business and lamented that tvs are low margin and that their services are higher margin. Okay, good, but then, like a day later, they came out and said we're thinking about manufacturing tvs. What was ural's reaction to that because you, you have been increasing your position in roku generally so roku's, an extremely interesting company.

From our perspective, the the the consumer court cutting as if anything, accelerated uh, it is likely that um there is going to remain an advertising monetized, um kind of professional tv service, a longish form tv service that that exists in the new landscape, and we think roku Is extremely well positioned for that on the tv side in in the player side, you should think of that as like, they are optimizing for customer acquisition in the most efficient way that they possibly can uh and so um. That's all you know they. None of that side of their business is designed to, nor should it take high margins out. It's just a question of how efficiently can you acquire customers via partnering via doing the individual um sticks, whatever mechanism you need to do so um on the um, but ultimately the the the pie that they're going for is that um video service should monetize, at some Rate per hour of of viewing, and so if you can drive and in particular advertising supportive video should monetize at a uh an exponential to how many total hours you can command of the overall advertising service landscape.

So it's likely to us that there's not multiple winners. In kind of the platform that delivers that advertising video service, that is, the alternative to old style, linear tv and that roku is well positioned for that uh and so the um. What what's interesting is like there's a conundrum for us where um viewing hours of linear tv have dropped, but actually had dollars, at least for the consulting agencies. The tracking don't seem to have followed yet so we don't know if they're being mismeasured now or there's just like so much inertia and lag in the in the ad buying space that that it is to come uh.

And so it seems obvious to us that the direction it's gon na go is is a service like roku's and also, if you think about what's happened with with apple and idfa, where apple's basically like thrown sand in the eyes of facebook and everybody else. So they can no longer like you know they can't they can't tell that their ad delivers to an actual transaction. At the end, this makes uh kind of roku's product offering actually more interesting and attractive because they have an alternative mechanism by which to track those consumers and can potentially compress right up against um the actual transaction side of things within the context of their platform. What about youtube? Uh, you know kathy has said before that apple and google are more like cash parks for her, but youtube's, crushing it.

I mean high margins, massive growth, any thoughts on it. Yeah i mean alphabet, is a obviously an extremely successful and interesting company. I think that there is always a question within the alphabet structure of how um how well they can incentivize people outside of core search to actually do aggressive work for the company, like waymo, was ahead of the field in robo taxi, and i don't know if they've Stopped making progress so they've, just you know they. They certainly haven't aggressively commercialized right and so they're, either up against a technological barrier, or they can't figure out how to motivate the talent to push uh, and you know they bought nest, which is like an interesting product, and then they ripped all the software engineers.

Out of it, and then you know so i think their perfect strategy has been pretty sclerotic in terms of enabling other things outside of core search to actually work as good businesses. Now youtube is, you know, a good business. It's super interesting uh and it's a question of one like does does is how much capital inefficiency is there inside an alphabet given kind of like those motivation, concerns um, yeah and uh? You know for us, as as managers, it's very easy for people to get exposure to alphabet. There's all kinds of you know.

Alphabet like the fangs are something like 40 of them and so uh, you know uh. We uh look for intangible assets that are more misunderstood. Uh, rather than um, something that you know it's you know people are heavily focused on and probably understand, more completely the value of the intangible assets they have yeah.