The AI Stock CRASH is Coming!

It seems like that Nvidia is selling a lot of GPUs, but it doesn't really look great on how they are going to get their money. So, they're they're not just selling them today. They are selling them, you know, also the the ones that they have uh what is it called like the orders that they have, right? They are selling the GPUs that they are going to produce next next year. And it's not very clear how OpenAI is going to pay because when we talk about like Google and Facebook, they have like actual money. They have a lot of money and they where do they have the money from? They have it from ads. So these are businesses that have been that they are proven that they are making money. They've been making money. But with Open AI, they haven't really made money. They are constantly losing money and it's just that we are we're basically saying is this a cycle that can go on forever or is this going to crash sometime sometime or or maybe it's not a cycle it's not going to crash what is the alternative what is not the alternative what is the goal the goal is for them to be profitable and this is what I really can see like I have difficulty seeing like how open AI can be profitable from a 20 buck a month subscription. Here's the point. Not everybody is going to pay. I legitimately considering I'm considering um unsubscribing. The only reason I keep it because I have projects and folders and instructions, but it's not really difficult to even move that out of the OpenAI system and just have it somewhere else for free. and o open source is getting better and better and all the all of these like huge amounts of monies that we are talking about is going like I don't know the an actual percentage but just if I would say a number it's probably like 90% maybe even more than 90% of the money is spent on training the AI and the inference is apparently much cheaper so if the inference is cheaper It means that until when are we going to train AI will we hit a wall or not? We don't know. And the inference which is us using the compute of these servers um to to run the AI models that is going to go down and yeah that that is not going to make up for the 90% of the money that they are spending on the training. So wow so much to say. So much to say here, right? Yeah. So, and and this is where why I think Gemini 3 has kind of really hit a peak. Like I think Gemini 3 will be good enough and it's more than good enough. Sorry. For most people and I think we'll have marginal increases from this point onwards. I really do. And what's going to happen is that technology is going to get better for us to be able to run this stuff locally. Don't get me wrong, AI has changed the face of the planet, right? It's ch we cannot go back now to certainly people like us where I can use AI to do so much more um be much more productive because of AI. Computers and our hardware will naturally increase to the capacity where we can run AI locally on our phones, on our computer. There'll be no need for all of these streaming services. is I mean I like uh you know I like OpenAI's ambitions and they're really trying to find they're throwing so much out there. They're really trying to find a use case for AI. You can't you can't knock them for trying but I I honestly just think that AI will be a very very good counterpart for us to live our everyday lives and I think Gemini 3 is good enough for that. That's that's just something I'm feeling recently like saying all that we weren't going to go into it but naturally we just waffle so anthropic are the only are set to become profitable two years ahead of um open AI uh anthropic are set to become profitable in 2028 open AAI set to become profitable in 2030 30, which by the way, that sounds a hell of a lot small shorter time than what I thought it would be, right? You know, you're talking about OpenAI spending billions. They were both spending billions, but for Open AI specifically throwing so much money at stuff to become um profitable in 2030, which is, you know, five years time, it's crazy. You can kind of expect it a little bit from Claude just because they are very expensive. I think people talk about Claude being so expensive, but I actually think they're realistic. I think they are realistic. I think so many of these companies are taking a loss, whereas Claude are being smart about it, being like, "No, this is how much it actually cost." They probably still are taking a loss to be honest, but they're more realistic with it than Open Hour that's giving it to you for 20 bucks a month. like it's yeah it's it'll be an interesting five years. That's what the news has been saying this week about um about profitability. But I don't I just don't know. All of this sounds like AGI is not real or not uh around the corner what they I mean no one knows really what AGI is, right? You say it changes definition every week. So once it comes out the the normies people us techies are so like this is the thing whereas normies are just like what does AI stand for? Yeah but we know we know what AGI is like we have a clear definition for it. Uh I am upset that some altman is changing the definition to to make it less, you know, AGI, less of a less of a a pedestal, putting it less and less on a pedestal so that they can reach it, you know. Well, the the the whole point of AGI is being general intelligent. It's like it can do what human can do but in at the speed of AI, right? That's it. That's it. But um what some altman is reducing to is like these language modules that are prediction machines essentially. That's not Yeah, I I'll I'll know when I see AGI and I will let you know. You hear that folks when we don't need to worry about when AGI is because Bar will let us know. He is the authority on it. [laughter] Yeah. I'm I'm very I'm very picky on this because yeah um I talk about you know my philosopher friend with uh I I talk to him about you know consciousness and creativity and freedom and all of that and what it means. Yeah. A lot of this and I I hate it when some altman comes and says uh we have this thing around the corner. It's AGI. It's like no no not really. This is part of a larger conversation on my show, Command AI, which we stream live every single week. We discuss the news and all things related to AI in the world of design and web. Catch us next week and join in the banter.