저자: ElitheComputerGuy · 외국인
요약 보기
AI 자산 과열과 'first mover advantage' 허구성을 주장. OpenAI의 초대형 인프라 투자 전략이 투자자 자금 확보 목적이며 고객 가치와 어긋난다고 비판. Nvidia의 오픈 물리 AI 모델 공개는 하드웨어(Orin 등) 판매 전략이며, 테슬라는 AI 우위가 흔들리고 있다고 평가. 현대(보스턴다이내믹스)의 실제 공장 투입 휴머노이드 사례를 거론하며 투자 관심의 쏠림을 지적. 결론은 질문 던지기 형식으로 명확한 투자 예측은 없음.
다듬은 글
Welcome back. As you know, I am Eli, the computer guy. Here for your daily dose of the roars as we scream about the very stupid stuff going on in the world of technology. And we are only going to scream about the stuff going on in the world of technology because frankly, the wider world is just too insanely stupid at this point in time.
The only thing my brain can handle is talking about the world of technology. But before we get into it, before we get into it, I do have a few things to say. Um, Ukraine, uh, Russia, uh, Maduro, uh, Trump, Biden, and I have Fousy. I have Fousy bolded, circled, and underlined. Uh, so there you go. Apparently, apparently that's how you get the the video views in the modern world of YouTube is you talk about any of these particular subjects and uh and people pour into your comment section.
Anyways, that's a joke. That's a skit. That's not what's really on my uh my little clipboard here. We all we all know what is actually on my clipboard here. So, anyways, uh I do these videos in order to support Silicon Dojo, free hands-on technology education in Durham, North Carolina. Also coming back to Asheville, North Carolina.
We're doing a class tomorrow night on pushing AI to the edge using Raspberry Pies. That should be a pretty good class. We have many more classes coming down the pike if you can come to in-person classes. is take a look at silicon dojo.com and sign up for any classes that you're interested in. Do remember free is not free.
There's a donor box link down below. Again, that's why I'm sitting here screaming every day. So, if you want to throw some money into the project, that would be ever so lovely. And now with that, let's get to the real Let's get to the real story that has nothing to do with Fousy. Fousy Fousy has nothing to do with this story.
Maybe if I just drop it drop in Fousy like every like five sentences or something, it'll flag the YouTube ALGORITHM TO GO, "OH, ELI is talking about Fousy. We need we need to blast this to like 5 million viewers." But I want you to know I want you to know this video has nothing to do with Fousy at all.
At all. Have you heard that YouTube? Anyways, I think this is kind of an interesting story looking at the world of technology and the whole question of what what businesses will succeed at the end of the day and the fallacy of the concept of first mover advantage. Right? So one of the one of the big ideas over the past 10 or 15 years in the technology world is this idea of hyperscaling and the idea is to hyperscale startup companies so that they essentially monopolize an entire industrial segment uh and then they can get uh rent seeking for monopolizing that entire segment.
Right? You get it so big so fast the moat around your business is literally just size. nobody else can compete with you and then you just reap in the same profits uh forever. One of the things I've talked about for a while though is that many times uh first mover advantage just simply doesn't work. It's one of the we it's one of the weird fallacies in our industry.
It's like this weird fallacy where we have so many examples of where it just does not work and yet people are still hyperfocused on it. Again, it's it's one of the reasons why Open AI, you know, wants one point god knows how many trillion dollars of investment, right? The idea the idea is that if they can scale their infrastructure to such an insane degree, nobody else will be able to compete with them, then they will own quote unquote artificial intelligence.
Uh the problem is, as I've talked about before, that's just not how business works. It's very interesting right now, right? Anthropic has come out, the Anthropic CEO has come out and literally has been dissing OpenAI's uh hyperscaling model essentially saying, "Hey, if you use better data and you focus on your neural networks more, you can actually extract better results out of using far fewer resources than if you do OpenAI's model." It's interesting.
The Perplexity CEO came out and it's something that I would agree with. He was like, why why do you want to spend $1.4 $4 trillion in capex expenses for data centers when everybody essentially has a very powerful computer in their pocket already, right? Can your smartphone run a 840 billion parameter model?
Absolutely not. Can it run a 2 billion parameter model? Yeah. If you start start doing things like time shifting, if you start doing things like uh for like analyzing photographs, if it takes a minute to analyze a photograph on your computer, uh but the computer is just doing that basically 24 hours a day, you already own the computer.
Somebody already owns the computer. Why why are you going to build a data center for all of these services when they can run things locally? Uh basically, think about it. It's kind of like the JavaScript moment uh for the artificial intelligence world. Why have your servers processing uh the back-end programming language when you can simply send text?
You can send the entire script to the web browser and then the user takes all of that resource load. Right? So there's a lot of questions out there about how artificial intelligence uh will actually look uh at the end of the day. But but there's still this massive push again with this idea of first mover advantage hypers scale and then for whatever reason you own the market.
If we look in the past, we look at companies like Blackberry, right? We now in 2025, it's bizarre to me. It's weird. You can tell I'm getting to be an old guy at this point. Is is people people talk about iPhone being the first smartphone, the first smartphone created by Steve Jobs? And you're like, but it wasn't by a long shot.
Right. There were Blackberries. Uh there were Palm devices. Uh there were Motorola Q's. Right. Right. Microsoft had a had a mobile smart mobile operating system for a decade. Mo it's called Windows CE long before iOS came out but in popular consciousness now right smartphones were invented by Steve Jobs.
Again that's that's just dumb. If you look at Yahoo I remember when Yahoo was Yahoo and Yahoo was sold off for scrap, right? Nobody even thinks about Yahoo anymore. Everybody goes to Google. Uh you look at some of these companies again MySpace. MySpace was, you know, the company literally got sold as as a tax write off as a at a certain point in time.
And so one of the important things to understand is that these companies that seem very successful today that have very spit and polish CEOs that come out and say all the right things. The reality at the end of the day is through uh macro forces through macro forces a lot of times they will actually lose lose the fa they lose the fight lose the race at the end of the day.
And that's one of the things I've been talking about with this whole artificial intelligence race that's going on. Not necessarily just the concept between China and the United States, but also just the the the concept of of what is the right path to go down. Right? One of the things I've been talking about for a while is that there's this idea that the AI stack is what is called a mature stack.
A mature stack is we know what we need to do. There may be some things you can do around the edges for optimization and that type of thing, but you basically know how to how to build whatever it is you're going to do, right? If you're going to build a web app system in 2025, we all know how you're going to do it.
You're not going to reinvent the wheel, right? You might use Python and something like Django. You might use Apache and PHP and that type of thing, but basically web app development is an exceedingly mature stack at this point in time. The creativity is in what you create, not in how you create it. But if you look at artificial intelligence, the big thing to understand is that it's not it's a it's an exceedingly immature stack.
It's an immature stack to the degree that Sam Alman uh and the I don't even know them, right? and the CEO of Anthropic and the CEO of Perplexi and the CEO of Deepseek and the CEO of other companies, they lit they literally do not agree on what the damn stack should look like at the end of the day, which is right, basically the state of where we're at.
And so, one of the interesting things in this particular state is what you see is you have a lot of companies out there that make money doing other things, but for some reason, artificial intelligence has cap captured the imagination of the world, right? AI Again, it's not AI. To be clear, and I'm not trying to be pendantic here, artificial intelligence simply is not.
Artificial intelligence is a buzzphrase. It is marketing buzzwords so that Sam Alman can fleece your grandparents' retirement funds for $1.4 trillion, right? It is the way to scrape money out of your your grandparents' retirement funds so that when they're starving to death and freezing because they cannot pay the bills, Sam Alman will have enough blood money to buy his 5,000 foot yacht or whatever the it is these masters of the universe actually do, right?
And so [clears throat] one of the things to to be thinking about in this environment is that if you start looking at it, right? So AI is a buzz word. AI is a buzz phrase, but it is very very popular. Like everybody knows it. Everybody wants to be on the AI bandwagon. So, one of the issues you has is you have companies such as uh such as Meta who just don't really need to be in the AI space to any large degree, right?
They do need to be in the AI space to some degree for for communications and advertising, that type of thing. They don't really need they do they do not need to be a leader in the AI space. It just honestly doesn't make any sense. But right, Mark Zuckerberg, Mark Zuckerberg wants to be a leader in the technology world and AI is where all the cool stuff is at.
And so he's building all this AI stuff for some reason. And then he starts dropping open source models. Starts dropping, right? Llama. Llama 3 was good. Llama 4 I've heard is not good, right? But you have uh you have Meta just just putting uh models out in into into the world for people to use. So you have Anthropic and you have Open AAI trying to make money off of their models and Meta just dumps models for free.
Again, IBM, right? IBM, what was the last time you thought about IBM until until I started talking about IBM? IBM dumped their granite models, right? What how does IBM make their money? mainframes. Oh, you hear you hear mainframe. You're like, "All right, Eli, mainframes are so old and stupid." Yeah, Main frames also bringing a ton of money.
A hell of a lot of people are paying their mortgages because mainframes still exist, right? They do mainframe. They do they do like massive Fortune 500 corporations. They do government. They do mainframes. They do big ass You are not going to buy a laptop with an IBM sticker on it. You're probably not going to buy a service with an IBM logo on it.
Basically, they are selling to the big big big big big players, right? But again, they got they got they got to prove that they're on the cutting edge, too. they got to prove that they're keeping up with the kids. So, they came up with these Granite models. And again, I I actually love the Granite model.
The Gran Granite models are fabulous. Um, we're using them a lot for the projects for Silicon Dojo. And they just they just released those. They just released those again. Granite 3, two billion parameter model. Damn fine model. Runs on a Raspberry Pi 5 for Christ's sakes, right? Anyways, and they're just they're just giving this away for free.
And so, it's interesting. have a again you have Anthropic, you have OpenAI, you have these other companies, they're trying to sell their products when everybody else is dumping stuff onto the market for free. Where this becomes really interesting is when you start looking at companies like uh Tesla.
And so what's interesting with Tesla, what's really weird with Tesla is [clears throat] Elon Musk says whatever the hell is going through his brain at any particular point in time. And one of the interesting things with Tesla is that Tesla is not a car company. I have been informed. I've been informed that Tesla is not a car company.
It's not a car company. It's not an energy company. I guess it's not a robot company. It's like an AI company. I don't know. It's it's whatever kind of company will will get him more investment money cuz that's the most important thing in the technology world nowadays. It's not about actually shipping products that people want.
It's about coming out with buzzwords so investors will will hand you more cash. But think about this, right? Elon Elon Musk him. I am not reading between the lines. I'm not saying, well, if you understand the brain of an aspie, I say he came out and said it. He literally said it on video. He said that if Tesla does not perfect uh autonomous driving, autopilot, that Tesla is worthless.
Tesla as a company is worthless simply selling physical vehicles. The only real value they have is with AI or whatever. Again, basically the robotic brains and why this becomes very interesting as I've been talking about for a while with Tesla is Tesla was a first mover. It was their first mover uh in the electric car space.
So basically they were first mover in the United States. They came in, they built a good vehicle and nobody else wanted to compete with them. Again, I don't understand. I like Ford. I have a plug-in Ford Escape. My wife and I almost bought a Mache. U Ford I think creates very nice vehicles, very comfortable vehicles, the whole nine yards.
But even even Ford, Ford even at this point in time in 2025, Ford is backing off of electric electric electrified vehicles. Everybody like not only does Tesla have a first mover advantage, but all the people that should be their competitors are like like for some reason they they they just don't they just don't like they're they're allergic to electrification for whatever reason, right?
So anyways, Tesla has been able to to to keep keep that up with the whole electric car thing. But the issue is once we start coming to autonomous vehicles, once we start coming to robots, again, humanoid robots, they're Optimus robots, is that there's there's a metric ton of competition in that particular space.
Whimo Whimo has been providing rides for years at this point, has god knows how many vehicles on the road, is in numerous cities at this point, doing exactly what what autopilot was supposed to do or full self-driving was supposed to do years ago, right? And what you start seeing is you've got Zoom.
Zoom is another company that's out there. you have all the Chinese companies out there and one of the interesting things you see is the idea of the real value of a Tesla vehicle being autopilot is that that is already being degraded even though they had they had first mover advantage if you go back to whatever 2015 whenever full self-driving or or autopilot first dropped right they were a first mover back then at this point they are a lagger they to to to being kind about it they are a lagard and why this becomes very interesting we talk about macro forces and competitive forces and all that kind of stuff is that Nvidia now is coming out.
Nvidia is now coming out and saying, "Oh, oh, you know those models? Oh, you know those models that companies like uh like uh Tesla think are incredibly valuable? Yeah, we're we're just going to give them away for free." Like Nvidia is sitting there and going, "You know what? We can sell the I think Orurin these Orin boxes for like $2,000 a pop.
You know, if if these models are free, we can sell a ton of Orin boxes, right? The whole idea, you give you give away the razor and you sell the razor blade. You give away the models and then you sell the computers to run the models. And so this is coming from Nvidia's uh own news whatever Nvidia's own blog.
Nvidia releases new physical AI models as global partners unveil next generation robots. So the big thing that Nvidia is pushing into right now is the idea of these world models. So the concept with a world model is that the world model understands the world. So you can create robots to fold clothes.
You can create create robots to drive vehicles. You can create robots, I don't know, to put knees into the necks of black people. You know, you can create robots to do whatever it is that you have the the funding, the resources for them to do. And what's very curious about this is again the kind of question of where is the value for a company like Tesla if their theoretical value proposition is in their AI and the robotic stuff and there's also there's already a massive amount of competition in the humanoid robot space.
Uh and then all of these models start start dropping for free. Are are they actually in a good position? Uh let's see here. Nvidia today announced new open models, frameworks and AI infrastructure for physical AI and unveiled robots for every industry from global partners. So every industry they want to be in.
The new Nvidia technologies uh speed workflows across the entire robot development life cycle to accelerate the next wave of robotics including building generalist specialist robots that can quickly learn many tasks. Global industry leaders including Boston Dynamics, Caterpillar, Franco Robotics, Humanoid, LG Electronics, and Neuroo Robotics are using the Nvidia robotic stack to debut new AIdriven robots.
I think that's also one of the interesting things in the modern world of like AI and robots and all that kind of stuff is how certain companies certain companies seem to to suck all of the air out of the room and nobody else gets talked about. again the Teslas of the world or the open AIS of the world or that type of thing and then you basically don't hear about anybody else even though they're actually out there doing the work right if you look at Hyundai so Hyundai as I understand p purchased Boston Dynamics uh so they they are actually building humanoid robots those humanoid robots are actually going into factories the humanoid robots are built the humanoid [snorts] robots are proven the human robots are now they're now scaling up the production of the human robot humanoid robots to actually put them into useful service.
Now, think about that. When was the last time you actually heard about Hyundai and when was the last time you heard about Boston Dynamics actually doing this? Like a lot of the stuff with the Atlas robots nowadays, you'll see them doing tricks and things, but for for some for some reason, they don't talk about them doing useful tasks, right?
It's all about optimist robots, right? Optimus robots that aren't doing right now for some reason take up all the air in the conversation when other people's robots that are actually in factories that are actually doing work that are actually proving the point. Uh they don't they don't nobody sees them.
They're just they're just oh we we around here we only talk about aspirational technology. See, Eli, Eli, we we've decided that it is more important to invest in aspirational technology rather than technology that's actually doing the work. Ergo optimist is aspirational technology that gets all the investment money.
Oh, those Atlas robots that are actually in the factory right now. Yeah. See that? That's not aspir Eli, that's not aspirational. They're actually doing work. Who wants to invest in doing work? We are Americans. We invest in dreams, not in building foundations. Right? I think it's kind of curious there.
Uh, quote, "The chat GPT moment for robotics is here. Breakthroughs in physical AI models that understand the real world reason and plan actions are unlocking entirely new applications." And so, this is a very curious thing with what Nvidia has said as CES. So, CES is currently going on. This is what he said at the keynote apparently is what they're looking at is basically either the chat GPT moment for these physical AI models or the Android moment for these physical AI models.
The concept being is can can we build a robust stable and replicatable software stack that then some that other people can go out and they can use to build their stuff. So you think about this like with Android the Android operating system right? What does Google sell? Google Google sells search services.
Google sells all of your information. Google sells that kind of stuff. They don't really want to make money off of hardware. But the thing is is you have all of these uh these people that wanted to create smartphones, but they didn't have the resources to create their own operating system. So what does Google do?
Google creates an operating system, a good solid stable operating system. Gives away for quote unquote free for all of these other manufacturers to use. They install that Google operating system onto all of their devices and now Google is making hundreds of billions of dollars a year because people are typing in search into the uh the address bar and that's automatically going to Google, right?
Give build the oper build the thing that vendors need in order to sell more of your products. And so if you have Nvidia coming out and if Nvidia is sitting there looking at and saying look we're a hardware company, right? Realistically at the end of the day we are a hardware company. we need to sell more GPUs.
Well, they they don't they don't want the software stack to be monopolized, right? If only Tesla is buying hardware or only certain companies are buying hardware, that basically means uh Nvidia is trapped to only the people that are buying their hardware. If they put this tech stack down on the table that is free, open to use, robust, the whole nine yards, that now opens by a tremendous amount the number of vendors that can then come to them to then buy all of their GPUs and their hardware to hopefully for them get their get their valuation up to 10 trillion or whatever else.
So, I think this is very interesting. Uh Jensen Wong, founder and CEO of Nvidia. Uh quote, "Nvidia's full stack of jets and robotics processors, CUDA, Omniverse, and open physical AI models empowers our global ecosystem of partners to transform industries with AIdriven robotics." But yeah, I think it's I think it's a value valid argument.
Uh turning today's costly single task and hard to program machines into reasoning general specialist robots requires enormous capital and expertise to build foundation models. Yes, that's why Sam Alman wants trillions of dollars. Uh, Nvidia is building open models that allow developers to bypass resource intensive pre-training and focus on creating the next generation of AI robots and autonomous machines.
These new models all available on hugging face include. So, hugging face platform you can just simply download things. So, you can just download these models. Nvidia Cosmos Transfer 2.5 and Nvidia Cosmos Predict 2.5 open fully customizable world models that enable physically based synthetic data generation and robot policy evaluation and simulation for physical AI.
NVIDIA Cosmos Reason 2, an open reasoning vision language model that enables intelligent machines to see, understand, and act in the physical world like humans. Uh Nvidia Isaac Groot in 1.6 6 an open reason vision language uh action model purpose-built for humanoid robots that unlock unlock full body control and uses Nvidia's Cosmos Reason for better reasoning and contextual understanding.
So basically, they're just sitting here going, "Hey, look. Hey, you you know all this hardware that would cost you like a hundred billion dollars to create on your own. There you go. Have fun with it." Uh Frank Robotics, Neurobotics, and Humanoid are using Groot enabled workflows to simulate, train, and validate new behaviors for robots.
Salesforce is using Agent Force, Cosmos Reason, and the NVIDIA blueprint for video search and summarization to analyze video footage captured by its robots and reduce incident resolution times by 2x. Uh, robotics is now the fastest growing C category on hugging face where Nvidia's open models and data sets lead downloads among uh a surging open source community.
To bolster this community, Nvidia is working with hugging face to integrate open source Isaac and Groot technologies into the leading LA robot open source robotics framework, providing streamlined access to integrated software and hardware tools that accelerate end to-end development. This collaboration unites Nvidia's 2 million robotics developers with HuggingFac's global community of 13 million AI builders, group models, and Isaac Lab arena are now available in the uh the robot library for easy fine-tuning and evaluation.
So think about this again. Creating autonomous vehicles, creating autonomous robots, all that type of stuff. If you look back 15 years ago, that was basically science fiction. If you look back five or 10 years ago, that would take you a tremendous amount of resources uh in investment to be able to create.
And now Nvidia is like, here you go. Here you go. Remember us when you need to buy your GPUs. Uh providing a cost-effective high performance upgrade uh path for Nvidia Jets and orin customers. Uh, the new Nvidia Jetson uh T4000 module brings Nvidia Blackwell architecture to autonomous machines and general robotics for $1,999 at a,000 unit volume.
So, the idea they're trying to sell these Orin, this is the Orin hardware. Um, they're basically part of the the Jetson uh family. The idea of portable AI hardware. So, if you buy a thousand of these units, the idea is that you can put these units into your whatever robot, vehicle, whatever else for about $2,000 a pop.
Uh, it delivers 4x of performance of previous generation with,200 FP4 Top teraflops and 64 gigs of memory, all within a configurable 70watt envelope, ideal for energy constrained autonomy. Uh Nvidia IGX Thor, which will be available later this month, extends robotics to the industrial edge, offering high performance AI computing with enterprise software support and functional safety.
Archer is using IGX Thor to bring AI to aviation, advancing critical capabilities and aircraft safety, aircraft integration and autonomy ready systems. In addition, Caterpillar is expanding its collaboration with Nvidia to bring advanced AI and autonomy to equipment and job sites in construction and mining.
So basically the idea here is Nvidia is dropping all of these models basically AI models used for building robotics under the world for absolutely free. Go and use them as you see fit as long as you buy their their GPUs and networking equipment on all that kind of stuff. And it becomes very interesting again for a company like Tesla, right?
Elon M Elon Musk has stated Tesla [snorts] is worthless without autonomous vehicles. It has zero value without autonomous vehicles. And now when he said it, he said that a couple years ago and the idea was the idea was that they will get their autonomous vehicles to market first and therefore they would then own the autonomous vehicle market.
So our vehicles are worthless without without autonomy. We are going to get to the market first. We are going to hypers scale. we're going to take over the market and then that's where it's going to be all worth gazillions of dollars. The issue is Elon Musk does whatever the it is Elon Musk does, right?
It's the sad thing the sad thing with Elon Musk is he does a lot of very impressive things. He does amazing things. If he just talked about the amazing things he's done or at least will get finished in the next couple of months, it would be great. That would be great. But no, no always got to be like talking about this There just not it's just not going to happen, right?
And so the problem is he started talking about this and now and now we're getting the cyber cabs. Supposedly cyber cabs are coming out uh in Austin. Now they're being seen in Austin. God knows how that's actually going to operate. But think about this, right? I if he said Tesla is worthless without autonomous driving, what does it say about the value of Tesla if by the time they got autonomous driving out into the public that autonomous driving is is essentially basically a free a free product, right?
Think about this with uh with with Netscape Navigator. I've talked about Netscape before, right? Netscape, they came out with a web browser way back in the in the '9s and they used to sell it for 40 bucks a pop. In the early '9s, they used to sell it for 40 bucks a pop. So that'd be like $120 in today, right?
Imag imagine like having to pay a hundred bucks or something just to get to a web browser, right? But but it was it was valuable. It was important. People needed it. They understand it. So the value proposition, it was sexy and the whole nine yards, right? And then then Microsoft comes out and God bless Microsoft.
Microsoft does what we do best in this industry and they just [snorts] they just take Netscape out back do some very nasty things to its bung hole. Right? They they come out and they're like we're just going we're just going to put Internet Explorer into into Windows for free. For free. So you can get all the power of a web browser for free or you can pay 40 bucks.
you can still pay 40 bucks for Netscape Navigator. And everybody used Internet Explorer and Netscape Navigator died. A horrible, god- aful dab. Why? Because Microsoft wasn't in the business of selling web browsers. To them, web browser is just another crappy piece of crappy software you throw onto the system.
But that allowed them to sell more operating systems, right? I I remember the days pre Windows 95 and post Windows 95. And here's the thing, many of us own computers before Windows 95. None of us knew what the to do with them. They were the they were the prettiest little um like fashion accessories.
Like you you would go into people's houses and you could tell how well their parents were doing. Like when I had friends, how well their parents were doing because parents that were doing well would have a computer in the corner somewhere. Now, to be clear, nobody actually ever used the Nobody [snorts] used it.
Like maybe maybe you might write a term paper on it or something like that, but that was it. Like we we again kind of like with AI, we were all told that computers of the future. We didn't actually know what the hell to do with the damn things, right? So everybody had they had their computer that collected dust.
Remember the old uh the dust covers for computers? Like we had dust covers for computers. And the reason you had dust covers for computers because it would be so long between the times that you would actually use the computer, you'd want to put a dust cover over it just to make sure it didn't get any dust inside.
That shit's real, right? Anyways, Microsoft comes out, right? You got Internet Explorer. All of a sudden, Internet, that's the thing. Computers, people didn't really know what the to do with computers. Computers, it's a future. We don't really know what to do with it. Internet Explorer. The Internet, people understand.
Email, they understand. Websites, they understand. The rest of this, they understand. So, basically, Microsoft is in the business of selling operating systems. They suction cup on this free piece of uh basically this free web browser. So now anybody that gets a Windows PC can now get on the internet that sells a crap ton of operating systems and Netscape Netscape basically dies.
And that's where it's kind of curious to see what's going on with this whole AI world right now and really wonder about who the win who the actual winners and who the losers are going to be because you have so much of this this weird weird stuff going on. Again, the the the bizarre thing in the tech world where where there should be technology partners, right?
Microsoft and Netscape should in a sane world, they would have worked together, but it's our [snorts] industry. It's Bill Gates, right? So, anyways, Bill Gates crushed Bill Gates crushed Netcape because it's fun. Cuz it's fun, right? And again, it's interesting like you you look at these companies where like you look at Amazon, right?
Am Amazon literally competes with its own vendors, right? So when you go and you start selling your if you're a manufacturer, you start selling your products on Amazon, Amazon starts being able to collect a tremendous amount of data about how people interact with your products, what they're interested in, uh the price points they buy things for, the whole nine yards.
And one of the big problems with Amazon is Amazon then goes out and literally creates generic versions of their vendor's products and sells them for cheaper, right? That's that's that's so up. That is so up. But it's true. That's that's that's just what we do. Again, and you look at like Meta. Meta is going out there, they're creating the AI thing, and then they they just give away they just give away their model for free because they're they're not in the model business.
They're in the the social media business. And so, what's curious here is like with uh with Nvidia, right? They're they're in the the sell hardware business, right? They need to sell more hardware. And I think one of the big issues that they're realizing is is that they're in a very bad point. Like it sounds weird to say, Eli, they're a $5 trillion company.
How can you say that they're in a bad position? Because they're a $5 trillion company, right? Investors investors don't want to pay for a $5 trillion company and have it being worth $5 trillion next year. $5 trillion company needs to get to 10 trillion somehow. Needs [snorts] to get to 10 trillion somehow.
And they they look around. They look around. how they look at the people buying their hardware and the rest of it. Are the path as it's been does that get them to selling enough hardware to get them to to 10 trillion? Probably not. So what do they do? They start giving they start giving away all the software this very very valuable software that would be incredibly expensive for anybody else to produce.
They start giving that away so that people buy more of their hardware and that is where they actually make money. And that's the big question in that environment. What happens to Open AI in that environment? What happens to Tesla in that environment? What happens to these other AI companies? And and as as I've said before, I think they become the MySpace and the Yahoos and the Blackberries.
Uh and they basically become the the footnote of of what was until we got to, you know, wherever it is that we are now. again. You know, I talk about things like Tesla and people laugh at me for whatever reason uh on Tesla. But you look at Intel right now, [snorts] like what the happened to Intel? Seriously, like Intel Intel owned uh the the CPU market, right?
They they owned the CPU market. They they they should be the front runner for the next hundred years until they dorked out. And now now God knows what like look at Intel. Like Intel is is is this generation's current Yahoo. you just look at it, you're like, why why why is it going so badly? I just don't understand.
Um, and I think that might be the future for uh for Tesla and the rest of these companies. So, anyway, so what do you think about this? What do you think about Nvidia dropping uh these AI world models and basically these open- source models uh for creating very powerful uh very valuable uh autonomous robotic systems?
What do you think this does to companies like Tesla who have literally stated they have no value simply selling hardware? The only real value they have is in the intellectual property of their AI and now AI is basically being given away for free. What do you what do you think about this again with the hyper the hyperscalers in the AI world trying to do the whole they they will scale to such a massive degree that they think they will own they will monopolize the market and that and that the plan seems to be failing in real time.
I mean think about this. I deal with artificial intelligence every single day. I'm on chatbt every single day uh for these projects for Silicon Dojo. I'm on those every single day and think about again not you know I talk about that with with thought leaders my thought leader and with CEOs decision makers or whatever like don't listen to what they say on specific sound bites right because sound bites are just sound bites look about what they're talking about every day look about where they're investing their time and energy and resources and that type of thing and think about this for a second I use catchy PT every single [snorts] day right I use it every single Okay.
But how much do you hear me talking about it? How much do you ever hear me being actually excited about CHPT? On the other hand, granite model, granite model I also use every single day from IBM. And I don't shut the up up about IBM's granite model. And the reason is is because IBM's granite model is I find actually be interesting and useful and paradigm shifting.
And CAT KPT's 5.2 2 model is just a pain in the ass at this point. [snorts] Like it works. It does what I need it to do, but it's quirky. Like like all of a sudden like you'll notice like my featured images, my fe like I had a nice style. I had a nice style set up for featured images. I like that style.
Uh and then Chad GPT changed something and now all of my images have an entirely different style. And again, for me, I don't really care that much about the featured images. It's just AI slop anyway. So yeah, whatever. But think about that, right? I I had an entire workflow set up for for Chad GPT. They they make a change to their system, all all of a sudden the output is massively different.
Is is that what people actually want to buy into? Again, whereas I use, you know, Granite or whatever else, I use I run Granite locally. It's going to do exactly what I tell it to do. I'm going to get the exact same results over and over again. uh the whole nine yards. So that's where it's like, yeah, what is what is the real value proposition there?
You know, it's like with uh with Sam Alman, right, when uh Kim and I 3 Pro came out and he did the red alert, right? They're they're on red alert over over at OpenAI and it's and it's fascinating, right? So they they had some kind of advertising system that they were dealing with. They had agents that they were creating, right?
They they had like five they had like five of these projects to extend the capability of the model, right? So with LLMs or models, you have the model and basically then there's extensions capability. So so agents allow you to I don't know order Uber or something else. They had all these extens and one of the interesting things is when they went to red alert basically he shut down all the people working on these extensions and then hyper hyperfocused on the found the frontier model itself.
And again that's one of my that's one of my big arguments here is I don't think the frontier model at this point in time is actually the really valuable thing. I really like in my heart of hearts I don't. But the issue they run in for these technology companies is it costs a lot. It costs a lot of money to improve a frontier model.
It doesn't cost very much money to create an agent, right? Databases, databases, web apps, that kind of is pretty inexpensive, right? Whereas frontier model development, again, that's what cost you a trillion dollars. So here's here's the issue, right? I if if if Open AI wants to get, you know, $1.4 4 trillion in capex expenses.
He wants $7 trillion in order to create processors and the whole nine yards. One of the big interesting questions is with the path for OpenAI, can open AI take the less expensive path, right? They're already they're at a $500 billion valuation now. They're trying to get to an $830 billion valuation with this next funding round and they want to get to a trillion dollars when they IPO.
What what what happens if they find out that the future the future of AI is creating a shitty shitty little web app agents that have a relational database backend, right? No, nobody's going to spend trillions of dollars on shitty little AI agents have a relational database backend, right? So, it makes it makes more it makes more sense if you're going after investor money to try to to try to go for that that more expensive product path because that proves that you need more money, right?
You don't need GPUs in order to create AI agents. You need GPUs in order to uh to create the next frontier model. And so again, that's going to become a very interesting thing is a as as these companies, as these leaders, as these CEOs, as they start getting caught in this position of the thing that you can sell to investors is diametrically opposed to the thing that you can sell to customers.
And the big problem we've had in the tech world for the past decade or more is that investors are exponentially more important to these companies than customers. [sighs] So what happens as more and more free open-source easy to access AI technology uh comes into the marketplace at the exact same time companies like Tesla and companies like Open AI are trying to guzzle down all the investor money that they can get.
Basically these these old school AI companies are going after the investor money. These other companies are going after the customer money. who wins in business in 2030 or 2035. So anyways, I don't know some things to ponder, some things to think about. What do you think about this with Nvidia releasing their models?
Do you think this is the future? Do you think Nvidia is just simply going to own artificial intelligence? What do you think about this again with open AAI being worth theoretically being worth $500 million? They want to go to a trillion dollar valuation to IPO. At the exact same time, like every other day, somebody is dropping a new model.
And uh and they all work pretty damn good. What do you think about this with Tesla? Right. The the whole argument with Tesla is right, it's not a hardware company. Hardware companies are worthless. Per Elon Musk, hardware companies are worthless. [sighs and gasps] So, so what what happens when they when they have the fourth best uh AI models for their hardware?
Does that does that make them a, you know, $ 1.5 trillion company or whatever they currently are at? Put your thoughts put your thoughts uh down below. Do remember to do these videos in order to support Silicon Dojo, authoritless gatekeepers, free to use hands-on technology technology education that empowers you to do whatever the hell it is you want to do in Durham, North Carolina.
Also coming back to Asheville, North Carolina pretty soon. We're having a class tomorrow. tomorrow. Tomorrow, tomorrow on pushing AI to the edge with Raspberry Pies. It's pretty cool. Going to show off a little autonomous robot. Uh doing all kinds of cool things with AI and Raspberry Pi. So, we're doing that class tomorrow in Durham, place called American Underground.
On the 17th, we're doing a full day class on introduction to Python programming for artificial intelligence projects. So, that's going to take you everywhere from understanding what a variable [clears throat] is to loops to function calls all the way up to building small little AI apps. That's going to be a full uh day course coming up on the 17th.
[clears throat] And then on the 21st, we're having a class on introduction to programming uh in an AI world, that type of thing. Anyways, that's going to be at American Underground 2. And uh yeah, that's all I've got. I've got to got to finish up. Again, instru instructors can be procrastinators just the way students can be.
I've got to got to finish up the final uh material uh for the class tomorrow and print out god knows how many pages of that workbook. So with that, see y'all
Comments 0