everyone should have ads and it's making cars safer and that's great but what unlocks the opportunity is the human not paying attention if you can't do that then you're not fundamentally changing the business model and you know that's kind of what we see with Tesla today with people forecasting it and where we think there's this huge gap you know a lot of people still forecast it and they say okay how many people are going to subscribe to FSD oh FSD gets better so they can now charge you know $112,000 and instead of 10% of the fleet now 20% of the will do that and it's marginal creative because it's software I would say that's kind of the status quo modeling of it whereas what Tasha has done and I think is the right way for people and the way that we'll see more people do it as robotaxi launches and beats people over the head with the idea that this is real is the per mile unit economics and that is a business model transformation right it's like humans are expensive I think that's the first point right over 10 years you know what is a human in manufacturing cost that's present value of $550,000 and that's not including turnover or searching for someone to hire so again to the broad point if you have a robot that can do what a human can do it's a no-brainer right it's like it you're not going to even if the robot costs $250,000 it's fine but I think the reality is where are we on this chart we're in this bottom left corner what is the productivity uplift of a robot this is where the math is harder for people to I I think grock uh intuitively which is what does picking and placing mean it's like what does a dishwasher in our uh house like what is that productivity uplift relative to being a human in a house value investors look for stocks with steady gains growth investors on the other hand are willing to bet on Visionary companies for future growth many investors in Tesla see it as not just an auto company but one that will transform energy autonomy and robots today we get to speak to Sam korus the director of research for autonomous technology and Robotics at AR invest Sam will share his thoughts on his incredible research how will Tesla focus on $115,000 purpose-built Cyber cabs and reduce cost per mile compare with its competitors what will the robot landscape look like and will Tesla dominate Arc invest is an investment management firm founded by Cathy wood the company's known for its focus on disruptive innovation Arc offers a multitude of investment vehicles that provide investors with exposure to these Innovative sectors including ETFs and Adventure fund we'll also be joined by Larry Goldberg A Serial entrepreneur and longtime Tesla analyst welcome Sam welcome Larry thank you for joining me thanks good to be here Sam I wanted to have this conversation with you because the work you've done for Ark invest and this big Ideas report that you guys shared uh is pretty significant um I think Larry n and many others think that the research you guys do is topnotch there's nothing else there I'm a big fan of disruptive innovation like looking for where Trends are headed and when they do hit they will disrupt entire markets so your big Ideas report I want to ask you like where do you guys see Tesla I know you guys are big investors in Tesla you your previous research on it has been dead on and curious to see where you're at now what is your thesis on thesa and uh maybe you can start with this explaining what this is sure you say you say me but first got to give a shout out it's uh the whole team so of course Tasha Tasha doing tremendous work on the autonomous side of things uh Daniel and AOS as well um really make it all possible to to pull together Tesla and the reason you know you need so many people Herbert ties in exactly to this right you see that first section convergence that really is the story of Big Ideas right everything that's happening right now the advancements in AI energy storage all of these things are uh catalyzing and being catalyzed by each other and so Tesla has been and continues to be one of our largest Holdings and it really is the story of convergence right you have the robo taxi side which is a huge portion of you know the value we attribute to it in our open source model you can see 90% of that is being driven by by uh Robo taxi then you have the energy storage side where Tesla's seeing tremendous growth uh you've got AI which is powering all of it as well it's really really is pulling everything together in a remarkable way so you see Tesla right that your thesis in Tesla is that multiple of these Innovations is all coming together it's going to converge um but right now your your forecast for Tesla is primarily built on the robo taxi opportunity when do you guys when do you think you'll include the Optimus um part of it into the valuation forecast yeah so that's this is a big big debate I would say uh not just inside of Arc but uh broadly as well and so you know we integrated it very minimally in last year's open source model uh really minimal impact just because of the types of assumptions that we were making even though we recognize that this could be a huge opportunity uh since then Elon has come out and said you know maybe 10,000 humanoid robots this year and then you know scale by 10x each year and then you've got a you know crazy Workforce of robots um I think with humanoid the way that we're thinking about it is let's see the execution and roll out of autonomous and then use that really as a guidepost for humanoid robots and the somewhat of the thinking behind this is a humanoid robot and the unconstrained world that it operates in is you know likely a thousand maybe even 10,000x more complicated than a autonomous car and what's amazing and I think what's really shifted with respect to robotics is that you know you go back 3 five years everyone assumed that this was going to be Hardware Limited right you don't have the actuators um or the sensors to make this possible now it's very clear this is software defined and software is moving extremely quickly uh but it does seem like autonomous vehicles needs to happen and needs to start scaling um as kind of this uh Milestone to hit as you have humanoid robots start to enter The Fray yeah I I think that's actually fair right so the I was telling you earlier that you guys have done tremendous research right and you've done deep Dives in Robo taxi and we're going to I think we should do that conversation first but doesn't stop you guys from doing deep dive research in robots but you're just saying well I GNA we're gonna think about the forecasting and one do we include in the forecast Let's uh take a beat that's fair I think that's fine as long as you know robots are coming and you do better than everyone else could I just raise a issue here well a question really not an issue so in your Venture fund you've invested in figure and I see recently in abtronic which I'm delighted to see since I'm an investor there too um it just seems a little odd that you're already investing in these two companies and yet not not recognizing the potential for Optimus in Tesla just well I you know I'd say you can look at the valuation of those two companies and it's like incremental progress moves those uh by a significant factor whereas uh a company as large as Tesla you know doing all of these other things they could all make the exact same amount of progress and it could be a 10x for a very small company and it could be at the margin for a company size of of Tesla but I Herbert I don't want to mess up with the uh scheduled events here so it maybe as a a cliffhanger for the robot conversation later is uh you know is this hype or is it not I want both of your opinions later and we can we can discuss right that that's go that's going to be a good one because uh that's very interesting both Larry and I obviously know a lot about robots and you do too with your research i' be curious where you guys land we do need to address robot taxi first this is imminent you've got uh Tesla continuing to uh SK forast that they're going to launch a robo taxi service in June in Austin Tesla just got a one of the first permits required in California it's just still a human driven uh service but it's those it's a permit that they need to do so going back to your deck that uh you you guys helped put together of course you were one of the people that put this together talking about Robo taxis Eves continue to take share from Ice Vehicles U and again you're looking at it from the broadly that eventually the future is autonomous EVS let me pause there and ask you that question um Sam like you know we've just had uh people very concerned that uh you know zeker has announced that they've got they're going to be offering their adap Advanced a Das solution to all their cars for free so they're a small smaller play in um China then you got byd about two weeks ago saying this as well it seems to me that in my mind I see these guys as like it's now table Stakes you have to say you have autonomous uh otherwise you're just an EV maker and the future is autonomous vehicles uh what's your take on what's happening in the in that field I think that's the right take and everyone should have Adas and it's making car is safer and and that's great but what unlocks the opportunity is the human not paying attention and that's really if you can't do that then you're not fundamentally changing the business model and you know that's kind of what we see with Tesla today with people um forecasting it and where we think there's this huge gap you know a lot of people still forecast it and they say okay how many people are going to subscribe to FSD oh FSD gets better so they can now charge you know1 ,000 and instead of uh 10% of the fleet now 20% of the fleet will do that and it's marginal creative because it's software uh that's I would say that's kind of the status quo modeling of it whereas what Tasha has done and I think is the right way for people and the way that we'll see more people do it as robotaxi launches and you know beats people over the head with the idea that this is real is the per mile unit economics um and that is a business model transformation yeah let's get right to that then I have to kind of like love these slides by the way these are this is good so this is like you know like you had all these companies announcing that they're going to do autonomous commercial launch as you know Tesla started talking about it and all these other companies came out and said hey we're doing it too but who actually ended up doing it it's Google witho and Tesla everyone else they don't have any plans of course this is the Legacy auto just something to keep in mind everybody will make make announcements but that won't happen yeah that's a great slide from Tasha right it's everyone everyone critiques Tesla for being late when in reality you know they just haven't quit right it's you can be a quitter or or you can be late exactly and then why don't you explain this slide yeah this is the business model really transformation that uh Robo taxi enables for essentially 100 years the price of track traveling cost per mile of a personally owned vehicle has been the same uh and the robo taxi is this opportunity to continue this cost Decline and in doing so you open up a huge amount of demand yeah this is going to change everything um but yeah may maybe if we go to the uh to the energy cost decline in the the robo taxi itself I know that's a I think the most exciting with recent this one yeah this this this one here um Do You Want To You Herbert you can contextualize with the recent happenings and announcements because even this now is is slightly out of date given uh yeah let me let me see if I get this correct right this is the cool thing about you guys is that you are you are researching it you're predicting what's going to happen and then it actually turns out the things that you're saying turn out and so one of the things that uh we were hoping to hear was whether or not the the cost of the Cyber cab can actually be reduced to be as low as possible what is their cost per mile but also what is their battery and their efficiency based on the batteries and it's interesting that Tesla decided to go with the small battery like uh you know as much as just as small as possible but have the most efficiency so let's listen to what France and Lis just said recently in an interview with Sandy Monroe our cars last the road taxi can last the full without having to recharge too much so we got a lot of models going around you know trying to figure out like how like a typical Robo taxi we go pick up someone go to the next job whatever whatever and we don't and we think that we can do that with like a really quite a small battery pack it's under 50 kilowatt hours and still have over like close to 300 miles of real world range what do you think um to me this is uh exciting I think initially they came out and they said 250 miles now they're saying you know maybe just under 300 so maybe this is more in the 275 280 range um I can walk if you pull up the slide I can walk you through kind of how we think about it from a rights law perspective and just very high level so one the efficiency crazy you know not not unheard of this is somewhat on path with what we expected but I would say crazy relative to what uh other people thought so being able to do something like 5.5 miles per kilowatt hour is huge um and you can see this reflected uh in this table to the right where you know what are battery costs today certainly under $100 um in China there are reports of lfp batteries in that $50 range uh that I think that will fluctuate and come down but certainly there was uh an excess Supply as traditional automakers cut back their EV plans and there was a glut of batteries but you know we think over the next five to seven years that probably comes down to really in the $30 range we don't want to we don't want to blow people's minds too much and you know they're trying to they're trying to do this in the in the next two years not five to seven years and so you know for the launch of cyber cab we we do think that $50 per kilowatt hour for an lfp is reasonable and then if you hit that 5.5 miles per kilowatt hour um the whole battery pack system right it it could be pretty inconsequential relative to what you would expect you know that $2,300 uh and then very broadly speaking the drivetrain of a vehicle tends to be roughly 20% the cost of the total vehicle and so you know even when you look at what the Cyber cab is it's two seats a drivetrain and and a screen so and an ultra light body right so it could be feasible that something like that would be $115,000 at scale um it'll be interesting to see what they add in how luxurious they make the interior heated or cooled seats um other features you know safety is always going to be a priority both on the autonomous side but also the passive side uh but I think that this low cost really sets a incredible base for a robo taxi platform Sam though there I've heard the argument on the other side where people are saying the cost of the robo taxi is is important but it's not that important as your people say because it's just advertised over 10 15 years anyways so it doesn't matter that my wayo cost me 100,000 but Tesla cyber cab is 15,000 the end of the day you take the 100,000 you spread it over months and how much money you're making on a monthly basis it's won't really that P pack what's your thinking about that yeah and this again goes back to modeling that TSH has done where it it does matter I think more than people appreciate not just from The Upfront cost uh and scaling it right it's like how do you what is the business model to scale for Tesla are you scaling because customers are buying the car and then deploying it or if you're weo is that something you need to take onto your balance sheet right that's that in of itself itself is a differentiation and an interesting question but then on the maintenance and repair I think it's very fair to think of that as a percent of the cost of the vehicle uh if you've got expensive sensors and you have to hire people to fix it because it gets banged up and even though they're going to be autonomous that's not going to mean zero crashes it it will mean far far fewer uh but that'll be a factor and then the things to consider with a robo Taxi utilization and dead head miles and so it's not just you know over time but also what you can charge per mile because not every mile is going to be someone paying for it and so it does actually come into play more than people would expect but Sam to be to be realistic if you're looking at the difference between an investment of $115,000 and $100,000 to earn the same return it's pretty crazy to say that the it doesn't really matter because the margins are so great because the investment level is critical to the return so if you got something that's going to cost you six times as much to get the same return your investment is going to be one six of they're saying that they'll lease it and now you turn into a monthly fee and that monthly fee is more than well covered by the monthly Revenue that you're going to get back because the differential is going to be very significant the differ differential between $115,000 and $100,000 whether you pay the capital cost up front or you pay the lease rate that component of cost is going to be you know six times the difference six six and a half times the difference or six and 3/4 times the difference that's a huge number at scale that you can't I mean when you talk about at scale you can't ignore things like six times six and a half times the capital cost yeah Sam with this does Arc um does Arc believe that Tesla will own all cyber Caps or will they sell the Cyber caps so it does seem like to start they're going to own them and they want to have full control and understand the system completely which is very in line with what you would expect from Tesla SpaceX any Elon company out there um I think what we're currently thinking and discussing is you know the what does the demand curve look like over the course of the day right it it es and flows right you got commute times then you have downtime uh and then you've got commute time again and so it might make sense for Tesla to own a portion of the fleet that would satisfy uh base load demand and so those Vehicles would they'd own enough such that they're very highly utilized assets and then then rely on customers to own and operate fleets to meet those Peak demand and you know to me this is another one of those surprising thing one of the a common push back people say is oh but people won't lease their cars to the network and to me May it's like I'm always shocked by this because like right now if you're in an Uber someone's in that car leasing it to you and so it's like you're going to tell me that this person would not do the same thing but not have to be in the car with me it's like that's a that's a crazy stance I think right money will make you do all sorts of things you didn't think you would do once you realize what I can be making two three grand a month um I guess my point was this if Tesla can actually make a $115,000 robo taxi they won't need to sell it I mean the argument was if it's going to cost me $60,000 per car yeah okay you know the asset the hit on the assets too much for a company to absorb but if it's 15,000 they might just own it I I still think you can get I mean for them it's like it's still likely a better business model to sell it right because imagine you sell it the car is not going to be utilized as much so it doesn't make sense necessarily for them to hold an asset that's going to be utilized less than whatever their threshold is and then for an individual you know they can call it when they want it they can make the money from it um it's like having that flexible respon it's almost like selling it is like a battery on the grid right it's like a flexible load and it's good to have batteries on the grid and same way it's like you want this flexible response uh from individuals okay but clearly they'll start the whole project owning the cars right and then they'll make the decision based upon experience and and learning so on and so forth one one thought I had is that since they're starting the uh project with X with y's and threes that as they feed in robot taxes they'll own their own robot taxes but use wi's and 3es of owners on the you know who who want to feed their their wise and threes into the um into the the network so it gives them a lot of flexibility yeah I think I think consumer they will sell WIS and threes to Consumers and then for Fleet managers they might consider selling you know fleets of Robo taxi otherwise they'll keep all cyber Cavs just doesn't make sense to me but so let's get to the big slide this is the big one I think right is like what is the cost per mile that Tesla is going to be offering it and how's it compared to competitors you guys did great job here explaining this in detail you're estimating 30 40% lower than weo as one example of competitors tell us how you came up with that yeah again so this is this is Tasha's work here but this really comes into utilization and upfront cost of the vehicle um right you can see that on on just the different price buckets here uh weo obviously using lidar and a much higher cost type of vehicle that weighs on on what it can do at scale so to to what you were saying earlier this is kind of the the answer to that question does does it matter uh the cost of the vehicle uh certainly we think so depreciation right here yeah operation support seems to be lower for weo that you're giving them such a low amount which I believe that they're hugely dependent on remote operators and that black line should be bigger wider well this is that's at scale you can look the early launch is uh to the left gotcha okay gotcha so there there's that and then uh Herbert maybe the last one here is just the uh percent of Miles traveled on electric versus is that the next one next SL it might be further further up I want to get to that slide this one oh this is a big one yeah well that that's That's a classic uh this one and so this is Daniel Maguire did did a lot of work on this one as well as a cost and the point here really just to be made is people are still so focused on electric vehicles and you say okay you're selling electric vehicles into a massive Fleet of vehicles and so what share of miles are driven that are electric is it's a slow transition but if you start to launch an autonomous electric vehicle Fleet this really starts to take off because and this is what Elon says in a lot of these uh calls right it's like if you go from a car that's utilized 5% of the time to 50% of the time well that's a 10 X increase in utilization and so you're starting to eat up more electric miles um than would otherwise be anticipated yeah that that shows you the commitment that elon's going to do because he wants to move it as fast as possible and the fastest way is actually autonomy that's how you're going to get at least five times you agree with that guess five times more than a regular one car you're saying utilization MH yeah yeah I think potentially even more than that but start there for for those that may not follow in detail I mean what we're saying is that a robot taxi is going to be on the road five times off five times longer per day than a than a car that's owned for personal use which sits in the garage for most of the the day or sits at the at the point of work or point of point of shopping or wherever it is sits parked most of the day and is driven only for a very small proportion of the day whereas a robot taxi is highly likely to be used much more a much larger proportion of the day and that's the real difference right so we so the theory is we'll need a lot fewer vehicles in the future and put a lot fewer vehicles on the road or will people use transport more often I mean you know very often these things the the utilization increases to the available capacity I got two slides here I really want to get to your opinion on I've been asking this question to everybody I've interviewed and now I get to ask you directly this is great this is so good so this one here right so um what you guys have done here is your you're forecasting when you think that Tesla will be able to have their you know software the full soft driving to be as safe as a human and what you've done by reading this is that you're basically using the uh compute as the proxy the more compute that Tesla is able to have the more likely that they're going to hit and you've got two Milestones here the 10,000 mil between critical intervention is equivalent to when one human's driving for one year worth and that's the a you know about 10,000 miles that there might be one critical intervention for one human driving you know typical one year's worth but then when you get to 700,000 that's when it's as safe as a human as safe as the average human out there and so you're saying well we expect it when Tesla is able to get to 190,000 h100 equivalents and you're thinking maybe by October of this year that's when we get to that that number um so I just ask everybody do you think that that's fair basing on comp I do uh some people think that this is kind of you know simplistic but I think it's the right it's the right one um tell me what your how you guys came up with this yeah I think it it is fair but I also don't want to I'm I'm Gonna Leave It Cliffhanger for your conversation if you if you speak to Tasha because this is I don't want to I don't want to steal Tasha's Thunder here and this is uh Tasha and Brett have done tremendous work so I don't wna I don't want to tread on on their uh territory here but we you know very high level yes you know we do think this is a fair way to do it I think we've seen these scaling laws continueed to hold that's why we still see these companies investing uh tremendous amounts in compute to to do this Larry what are your thoughts yeah I'm aligned with uh these numbers I've studied these numbers quite carefully uh when they came out um I I think they're conservative to be honest um you know I've heard Brett explain Brett and Tasha discussed this in detail and explain how they came to this number and the reason I'm thinking that they're very um conservative is that we have taken the kind of general population of driver um in into account and um one of the things that is true about the average population of drivers is that's offset by the enormous um accident prone the the the enormous number of accidents inflicted by Young Drivers they really skew the balance very dramatically it it may be true given what we're seeing in Trends today that the young drivers are the ones that are the first adopters and so you know we may see a surprise in terms of accident rates if that if that holds I mean so many young people are not getting licenses are using Ubers and you know if that Trend holds and particularly if the prices drop it's going to I think explode I think we may see the new driver syndrome disappear soon and the also the older driver also accident prone people are probably not as serious accidents as the young people getting into but so I think the um the statistics may be skewed by that component I may be wrong so when I look at this first of all I I loved it I loved it a lot the reason is that it shows you that uh it's really now compute it's like based on compute will we get to critical miles per intervention the answer is yes if you increase compute and I think that that I can I agree with right because it's that's what Tesla's been able to show right more compute more time in compute by the way that uh allows the system to get better and better and better and then they can map it out they've been showing that based on your sh sharing what has happened in the last year so I like that they tied it to compute this idea that you need 190,000 and it won't be October that's where I'm going you guys think you're conservative and I'm like well that's you know Tesla's announ they said that they will get to this level by mid year so the question is what do they have access to more compute than we realize do you guys think that uh Tessa using the compute that's available in xai or they're two separate systems two two separate systems is my understanding I'm aligned with that yeah there's been reports and I know you can't trust reports but but there's reports that Tesla is using uh the Memphis supercomputers for FSD doesn't think so okay yeah yeah I I filled out my uh final four brackets so we'll see everybody else I think that may grock may be better at at predicting but it's interesting how grock has paed that because I've anyway separate subject but grock thinks that they're not Gro says that there that all that it can GA all that grock can gather indicates that they're not using that Tesla FSD is only using Tesla's uh supercomputers not XI supercomputer yeah and if you look at Tesla's um diagram of compute capacity which they did I think a year ago um I don't think attracts to where the supercomputer is in Memphis by any manner of means so you know it yeah anyway yeah regardless I mean you know what you guys are saying October that that's a good number for me it doesn't matter if it's let's say delayed because remember this is as safe as a human which means at any point a to any point B any human you know in the US let's say right so that's pretty significant and that you can reach that as soon as you get this and I'm not concerned that Tesla will not be able to get to this level of compute sometime so done deal that's right should should we uh raise the the humanoid robot question or or do you have a i one more I just want to go through this one and then let's go to human or robots but I just I think this is such a critical slide I think you guys did a great job here as well basically you're saying here that if you are competing with r Hill you want to be like an Uber Market that's $134 billion Market but as soon as you bring the price down to instead of a uber and lift they charge $2 to$ four dollar per mile if you can get it to a dollar per mile in the western markets you're already at you can grow the market to not just 134 it's almost like 8X to one trillion then if you get it 60 cents a mile you're now talking about different kinds of commuting and then it just keeps going to once you get to 25 cents a mile this is like all sorts of things that you haven't thought about it's A5 trillion dollar market right yeah and again this is this is a chart that Tasha has made and been updating uh for a long time now and I think that Herbert to your point this early Market is extremely potentially extremely profitable because you can look at Uber and lift today and you can see the amount of rides they're giving at the price point and then imagine if instead of it being a cost of paying a driver you just strip out that driver and they offer that price umbrella so you know you call an Uber it's $2 or you call Robo taxing it's a190 you choose the $190 option every single time but it costs you know a fraction because you're not having to Source drivers constantly uh that's that's really profound and it's it is at the Leading Edge of adoption right and not to mention the better experience that everybody's reporting in the wayo cars versus The Uber cars so that that is a an important one of the factors that I think is interesting but as yet unexplored is the impact of wides scale adoption of robot taxi on the roads in general but that's probably for another day have you guys started um planning that out researching that that the the impact of what happens when there's all Robo taxis yeah yeah yeah well there's one more traffic I think people say oh no everything's going to work perfectly and you're going to have cars flying and zipping it's like I think the reality is far more traffic and this transition period is going to be interesting too because in theory there's this point in time where it becomes uh exponentially more painful to be a human driver right because you don't necessarily mind being in traffic if you're in a robo taxi because I don't know you're on a call you're watching a movie you're doing whatever but if you're a human driver and you're in a traffic of Robo taxis that is certainly going to be infuriating and painful uh to live by so I think you know more more cars on the road leads to more traffic that creates some interesting opportunities for getting places quickly what do you do there I also think you know the short Hall flight has been something that we've discussed where you know Larry you're you're Mr road trip I'm sure you can draw the IUS of where it is much better to even even now driving with FSD and you're behind the wheel it's just a better experience than going through the airport having to worry that you know your overhead bin space is going to be taken up and then you wind up having to check a bag and then you're in the airport for an extra hour as opposed to hey here's my car and it's uh a breeze to go three four hours and that's that's a big Market as well I go 24 30 hours and it makes and I've just been on a air air trip to Austin and and St Pete and honestly I regret that I didn't take Drive instead so yeah but I will challenge one thing you may get more traffic more cars on the road but you're going to get much better traffic control and when you get better traffic control better driving you can put far more cars on the road road than we currently have the only Joker in the pack is whether we get FSD or we get autonomy on trucks because the trucks are the killer on in terms of traffic pattern but anyway that's I think that's an interesting debate too right because it's at least in the city so much traffic is caused by stopped vehicles I don't think that's going to change I think right and then it's like you get an autonomous car that's like delivering food or something right you know I think absolutely I think it's G to there's be a lot of complex externalities that will emerge as this rolls out your point about the cities is absolutely On Target but on the freeways behavior of the autonomous car I've seen it in my own car in my own autonomous car how well it behaves on the freeway and how I allow it to behave that and I don't start pass you know I don't want to get ahead and stuff like that I just let it do its thing it's a big Improvement in terms of traffic control and that will H that will en able cars to move to for the traffic to move faster anyway that's the subject for another day we can we can ble that one out later stock humanoid Bots next big topic so you um you know you you got we asked the question earlier when is Ark going to start including humanid bots in the price target for Tesla we're all been watching in the last year just an explosion of humanid bots everywhere this is part of your research as well these are the Bots every day we're seeing and and this this doesn't even include the you know hundreds of Chinese Bots that are out there they're just being able you know we just saw 1X demoed at a conference there and he they were using 1X was vacuuming was actually vacuuming using a vacuum cleaner we I think everybody here agrees that Tesla is the leader because of their manufacturing capabilities and so yeah curious what you're you're thinking about that then all right I'm going to go little little rant here so then cut cut off and and and uh we can go from there okay I think the math of humanoid robots breaks people's minds and so there's this dichotomy that is emerging because it's such a big opportunity and in order to be useful they have to do so little so it's like oh that's a no-brainer right it's like this is in in that slide you can see this is a$ 26 trillion per year opportunity potentially bigger right it's like if you replace all humans it's even bigger than that and so all you need to do is something and this is like a piece of this pie is tremendous and at the same time an autonomous car is in solution right it's doing every single part of the problem whereas a humanoid robot in order to be useful it doesn't even need to do the full solution right it's like you we did a slide on um professional dishwashers right and I think the dishwasher is an interesting thing it's one of the Staples of Automation in all of our homes and yet a professional dishwasher doesn't wash so many dishes of course they do right but it's like you look up what are the tasks that they have to do and washing dishes is one of a whole host of things and it's like we pay professional dishwashers $14 billion annually in the US despite the fact that we sell dishwashing machines right so it's like oh this is a huge automation that we've already solved but just kidding it's like we actually pay people more to do quote dishwashing than the actual Hardware itself and then if you look at what we do at home we all have dishwashers not everyone but a lot of people have dishwashers uh and if you value the time that is spent on still you know cleaning up after meals it's a $246 billion opportunity so it's like if you had if you had a robot at your home that all it did was clear the table and load the dishwasher it's this huge value and so you know it's like this is so big and so obvious and I think the hard part here is that it is hard it's like what does it actually mean to be able to do a single task it's much harder than I think people anticipate and so yes it's true like picking up a box and putting it somewhere easy and that's kind of this early application that we're seeing from I would say almost all of the humanoid robots companies it's like there's a huge huge Market just lifting a box and putting it down at the same time it's like just washing dishes is an incredibly complex thing to solve and and so it's like this is this is the balance of where we are and uh the difference between demo deployment and scaling and so that's that's where I think it's very difficult to to say where we are I think there's such excitement because software moving extremely quickly we should be excited that's absolutely true we should be excited huge opportunity and we should be excited because unlike autonomous driving it just has to do a single thing and where I'm hesitant and I think there's still proof to be done is okay let's let's see it like let's let see it Beyond demo in a factory um at scale what are your guys thoughts yeah so you're being C you're being cautious basically saying yes all is exciting yeah cautiously optimistic yeah yeah I think it's fair I I I guess my quick thing would would say this okay so um you you're I think you're obvious of course you guys are right you've done a ton of research but the hardware was the bottleneck and that and that no and that then it became once they were able to solve the actuators then they go now it's a software right we can't we don't have the brains and all of a sudden the software expanded to the point that they realized it can actually do certain tasks autonomously then it became the hardware again and it kind of continues to this battle and you heard Elon say in the last earnings call he said something like we need to have the tasks to scale to the number of bots we make so right now they can make thousands right they'll make a thousand or 2,000 this year year next year they're going to make 50,000 let's see if that happens or not but when they make more than a th000 2,000 5,000 what tasks are going to do well the good news is that I think the tasks for pick and place I think you agreed with me there's a huge number of bots you can use for those tasks let's say in a in a typical Tesla Factory they can do two to 5,000 Bots just doing pick and place things that is doable now I think personally so I think there's room for or let's say 5,000 per Factory let's say 100,000 200,000 a million Bots out there if you start selling it out there now when you're then by the time that they can make 100,000 or a million that's when you need to ask the question can they do more tasks and these tasks are much more complicated you realize that's a year or two years away well I think yeah I think the point you brought up is an interesting one too because so this was something we debated internally will Tesla sell these externally and you know it's like well no if they're if they're great then they're just going to keep them and use them but then Tesla's saying no you know we might do this and I think you're hitting the nail on the head with what you just said there which is this will be something interesting to see because if you can make 50,000 of them and they're really capable then in theory you can find uses for them and right do a crazy amount of things but but that's not what's going to happen if all it can do is pick and place there's crazy demand for that but probably not 50,000 pick and placers at Tesla and so the degree the degree to which they start selling externally is going to be reflective of you know limitations what is what is the actual software capability versus um yeah reality and you know the promise though then right it's like why humanoid at all is because it's generalizable so you as software update come out they should become more and more capable yeah I think I think that it's um that the it's going to be a it's going to be a while but you know they're going to be used for dirty dangerous and difficult DDD to 3D tasks at the outset um but start at the very simple level and then build up upwards I there are so many manufacturers and there are so many um sales groups looking for you use cases it's interesting that Boston Dynamics did not find a use case for a humanoid bot in all the time that they were you know building these very expensive Bots but I guess it's just because they were so expensive they tried to make you know for for their warehouses they made very very Warehouse specific complex Bots and they implemented those but um the the value add of the bot even for those 3D tasks is potentially so high you know I've done the numbers and the discount value of just one bot could be qu of a million dollars even on you know relatively low pay scale so there's driving there are a huge number of Manufacturers there's driving need in the marketplace and I think those those forces are going to come together to they're forcing fun it's going to happen once the bot can be manufactured and once the once the compute power gets to that but I think that we're far from solved on the compute power I think there's going to have to be a significant Edge compute functionality in each of the major locations where they used at scale anyway that's just some thoughts that I've had yeah maybe if you have the slide on that uh crossover point when yeah there look at you're Pro you're Pro but I'm a pro you're you are pro um right it's like humans are expensive I think that's the first point right over 10 years you know what is a human in manufacturing cost that's a present value of $550,000 and that's not including turnover or searching for someone to hire so again to the broad point if you have a ro OT that can do what a human can do mhm it's a no-brainer right it's like it you're not going to even if the robot costs $250,000 it's fine right it could cost $500,000 you say that's great this is amazing but I think the reality is uh to Larry what you were saying is where are we on this chart we're in this bottom left corner yeah where it's like what is the what is the productivity uplift of a robot and I think this is the hard this is where the math is harder for people to I think grock uh intuitively which is what does picking and placing mean it's like what does a dishwasher in our uh house like what is that productivity uplift relative to being a human in a house well I don't know it could be 0.05% productivity uplift depending on what you're doing and I I think you know what is lifting a box slowly or what does having to train a robot and I think this is a key piece of Robotics broadly is if the robot can come into your factory and just work huge that's like amazing because one of the biggest constraints on robots thus far is the fact that the robot itself doesn't cost that much but you need to pay a professional to then come in and set up your factory and then you need to pay someone to program it and then it's going to do the exact same thing over and over again and so what's what is exciting about has to be maintained don't don't underestimate maintenance and and and continuous training well there'll be continuous supervision I guess right and so it's like okay humanoid robot it is exciting because we're starting to see demos T operation it can learn a task in you know two days and a week whatever and then it can do it okay but can it come into a new Factory and do it or does it then require a week and you're like okay this is how you're going to set it up you need to change your process here all of that then is eating into the overall productivity uplift that this humanoid robot is doing and then it's okay it works and it can do it 95% of the time okay but then for that extra 5% that requires you do this process and you have these people watching and then okay you take all of this into account and then what is your productivity uplift is it 5% is it 10% is it 1% and that it's like this this is the thing that I'm watching most closely to say is this like the slide that Tasha has where it's I don't know 2015 and everyone says autonomous driving we're doing it in the next five years and here we are 10 years later and it's about to happen or is this something different so I mean Sam just just what I did is I I said there's no way the bot is going to come in without cost I've added lead in this tremendous amount of supervision charging maintenance and management and I've looked at the annual net saving at at the mean level of of what we pay people in these factories and and in warehouses and so on and I've discounted not at 10 years but five years at a 20% discount rate and that b is still worth quarter of a million dollars that is a lot of money even at the cost even at the B material cost that they're running at right now I mean they're they're running below 200,000 maybe above 100,000 but but even at the bill of materials cost before they're in mass production it it looks to me like there's an opportunity for early introduction and the earlier we introduce it the longer it's at job the longer the more work it does the in more intelligence going to get so I think I think we're at a much closer point to that magic moment to the to than than than you're than you're believing at this point that's just my view yeah yeah I agree with that too we've done a lot of work with CERN and we we priced in like having a manager human manager managing you know these Bots work three shifts not just one they and we we've worked it out that they can actually take the same brakes you know 15 minute battery charge 30 minute battery charge 50 minute battery charge just got human would and yet be able to work three shifts batteries are up updated human managers plus when you train one you've trained them all so you're right about the first shift maybe the first kind of thing but so for me I look at it based on the rate limiting step is how many Bots Tesla can make if they can make 5,000 this year and all it is is just picking place this year they'll need it all that's great next year they're saying 50,000 if all they can do is still pick and place like we just talked about it now 50,000 that's fine they'll sell it to other manufacturers for 50,000 for pick and place they're going to sell them all now the next year when they make 500,000 you're saying to me well maybe by that point they won't be able to do many they they just won't have that uh skill to be able to actually do a job yet Beyond pick and place and you've already you know maxed out the pick and place I don't agree I think that there will be many jobs that you can still find that is very limited that you can still be trained and still have that positive margin positive cost uh per per job per hour so I I think it's just a matter of building it and by that point that you know the compute and the tech the skill and ability for these Bots to do things that remember you said it too right the task is unlimited unlimited with with with EV they have one task right driving you from point A to point B and so the margins are really low and you it's hard for you to sell in a competitive market in Bots it's like you can say okay it can't do dishwashing yet well then let's do you know name one other task and all of a sudden do that task and then calculate how many of those you can replace so yeah I'm I'm very you know you see our investments I'm very optimistic yeah cautiously yeah yeah yeah I'm kind of between the two of you guys I think her's way out there I think what did I say that's what I say strike reality will strike very early in those marvelous plans on the other hand I don't I'm not as cautious I think as Sam I I see a path towards you know there aren't that many pick and place jobs in Tesla I've been through all their factories not all I've been to two of their factories in great detail there are not that many pick and place in the warehouse there are quite a few but I'm not not sure there's 25 5,000 give or take and particularly if it's working three shifts or even two and a half shifts that reduces it by by a third or a half so I I'm not sure there's that that things are going to scale quite as quickly as that let me just let's go to it okay they're going to make two to 3,000 Bots this year okay we all agree at this point that Tesla will be able to use that if they make them I don't agree with that necessarily listen I don't know if you've seen the video of the um star uh Starling Factory in Bastrop yeah y there are no human beings I mean literally no human beings I mean I've been there when the shift changes there are no human beings so I've been asking this question but if you look at mobile phone manufacturing there are thousands of human beings very very different that's very asked that question like I don't see any do you need human beings to create a bot you know does bot needs to make Bots and that's their answer well manufacturing them is the easy part I mean it's very difficult right now but it's the easy part in terms of the business plan hard part is getting them to do the job yeah and then that gets to another interesting point which is we believe right humanoid robots are going to create more automation right not it's not like oh you don't need all the specialized automation anymore it's like oh this actually leads to a lot of other things that you can automate and other small manufacturing companies that would never be able to automate that can now automate with specialized use cases um but then maybe you know I know we're we're running up on some time here it's like huge opportunity we've gone from Hardware to software constrained so software is going quickly and it's like okay so the and this is going to happen it's going to be a huge thing and now the question is how quickly does it happen yeah and and I think the bigger the big problem that people tend not to consider is the use case Perfection problem it's going to take a while to adjust and figure out how Bots and people work together I think anyway uh does Arc consider that Tesla might do more than just humanoid Bots um they've referred to quite a number of the Tesla Executives and AI team has referred to Bot animals and even Elon continuously every time he's on a interview he always says you know everybody will want to see 3PO and R2-D2 he's always talking about R2D2's too so you're you're asking if they're gonna make the uh scale BS we yeah yeah yeah different kinds of bots I think you you never know the the opportunity is so large for humanoid I'm imagine that they'll want to same thing with you know autonomous it's like why didn't they do a lot of things it's because well autonomous is the opportunity so don't get distracted when you've got the the thing ahead of you still to deliver on and I think that's true of autonomous they'll deliver and it'll be uh huge opportunity and then same thing for humanoid robot I don't think they're gonna distract themselves until they've really gotten that flywheel going yeah agreed but they will be doing it that's my point is that there's going to be so much more than just humanoid robots who always have positive um it's like I you know not a criticism no no no I think optimism's good I mean to be clear I think we're we're all agreeing that human abots is a huge opportunity we're just debating the timing and timing is going to be very lumpy so yeah it could be that next year the Bots don't have anything to do and they're not there yet not saying that that's going to be for sure happening just you know just remember where we were a year ago in January of last year nobody of so few people thought that we would be here today with how many bot companies what they're capable of doing or like you said they're demos we don't quite know yet Herbert I want to ask you a question I W it won't be the last question because we need to end on a high note but but but given given Larry's question just there what do you think is the most overhyped uh technology right now oh wow that's a very good question I think the most overhype would be Quantum Computing um that's the one that they they come out there and they make these presentations and you go holy cow if that's true and then when you dig a Little Deeper it sounds like that there's still kind of research and nobody's proved that it's actually working so I don't know that's fair that's a that's a good good take yeah I mean the thing about human and Bots is that um so so is it more so fusion fusion reactors are less or more than I have no idea you as this is the question for Arvest they're the ones that that's the research this you'll have to have Daniel Maguire on he's done he's done some very deep Dives on all things nuclear so he can he can share his that be great yeah yeah he's he's bloody good yeah yeah but I mean I think we probably agree here that AI is not overhyped and it's it's going to deliver more than people are expecting um and there it's actually at the very beginning of what it's able to do so I think we've got huge surprises to come in AI still in terms of the evolution of the technology huge surprises all right well let's talk back to Tesla AR invest um you guys are big Believers in Tesla what is your thinking right now or is your confidence even higher now as the months go by here as we get in closer to Robo taxi what do you got what kind of conversation you guys having in the in the office there yeah I think it's it's an interesting uh time it's always an interesting time what am what am I saying uh right it's like robotaxi is getting closer and closer to commercial launch this is the big opportunity simultaneously you've got so much noise and backlash against Tesla right you know domestic Terror attacks lighting cars on fire uh things of that nature and so it is it is interesting because yes it's like we do think they're going to launch everything we see is indicative of continued advancement on robotaxi and that you know you can look at the open source model that's where we see all of the value um and so yeah it's it's the perfect uh illustration of short-term versus long-term I would say and it's like yes shortterm there's certainly things that we did not expect to occur I don't think anyone really predicted the type of backlash or attacks against vehicles and uh you know this is the first quarter they launched uh new vehicle across the globe simultaneously so I think that's interesting of itself as an accomplishment but also it's getting publicized as a a negative because it is adding noise to the story of is there demand is there not demand Osborne effect all of this and I think you know in the short term Tesla continues to be a volatile and noisy story but I think if you take a step back and you look at the long-term narrative Arc and execution it's like the noise Fades away and it's pretty clear that they are executing on this longer term thesis perfect Sam you're very articulate I appreciate that so people can go to arkinventory um research rep or the 202 Big Ideas report and I saw here that you guys are releasing something new is called the investment opportunity report what is that yeah this is a it's like a cut of Big Ideas more tied into the fund side so the Big Ideas is more research driven and does not associate necessarily with specific um investment opportunities this is you're going to look at it and you'll say Okay I I get it but it it's more tied to the fund side as well so pairing it with the strategies and Investments that go alongside it gotta perfect thank you so you can follow Sam on X at sorus ARK Arc I love your account by the way you ask good questions um I didn't even get to chase you down a couple of points you made there like yeah is uh energy going to be you know we going to have an abundant energy or going to be scarce energy and that's a Hulk show we should do one day in the future thank you so much Sam appreciate your time thank you Larry yeah that was fun appreciate you guys bye everybody I've created a website that is the most comprehensive resource for the Tesla investor please check it out Simply go to my website at herbert.com [Music]