Is the AI Bubble About to Burst? Nvidia’s Earnings Reveal the Truth

Craig: everyone is asking if the ai hype train is finally about to derail. well, nvidia just dropped their latest financial numbers, and they have a $57,000,000,000 answer for us. if you hold tech stocks or use ai tools, you need to hear where all this money is actually going. nvidia just released a q three earnings report, and they completely blew past expectations with over 57,000,000,000 in revenue. this is headline news because nvidia is a shovel seller in the ai gold rush. their massive sales numbers prove that big tech isn't slowing down. in fact, companies are spending record amounts on chips to build their ai future. the report is effectively the industry is saying the bubble isn't bursting. we are just getting started. what do you think about this, john?

Jon: yeah. so this is crazy. and you were talking about their their spending unprecedented amounts. they're spending about $400,000,000,000 this year building ai infrastructure, and nvidia is the primary beneficiary. though, there's a there's a huge caveat here. okay? critics are worrying that this resembles the comm bubble because a lot of the deals that are happening involve what's called circular financing. so what that means is nvidia invest billions in ai companies like openai, who then use that money to buy nvidia chips. so it's like think about it like ford lending money to car buyers who can only purchase ford vehicles. so the question is, is this genuine demand driving a technological revolution or is it financial engineering inflating a dangerous bubble? my opinion, i think it's a mixture of the two. but we have to be very cautious here. the thing is, is that beyond gpu's though, nvidia, they're selling complete systems which include networking, software frameworks, and cloud services. so the integrated approach that they're doing, that creates substantial switching costs for customers, and that explains why these gross margins remain robust despite intense competition. so the thing is here is i think, yes, we are in a bubble. i think the circular financing or the circular moving around of money is is adding to that. how big of a bubble is gonna burst? i don't know. what do you think?

Craig: well, i've been talking about the bubble for a while. i just think we're in a bubble. it's just not ready to burst. i've talked to i can't remember which investor it was, but they invest in a lot of ai companies. and i was talking to them a couple of weeks ago, and they were surprised when i said it was a race to the bottom with these ai companies. and, you know, meaning that, what was $200 a month is $20 a month, and they're all kinda going after each other's segment. so i need better powerful cpus so that my lower priced offering can beat your offering, and it becomes this race to the bottom. and then they're gonna eat itself, like, things like cursor and lovable, and, eventually, they'll be gone because the big players will just have those tools built in. and they do. well, they they do, to a certain sense, are just not as good for, a bunch of circumstances. yeah. depending on how you approach it. so, you know, they anticipate 3,000,000,000,000 to 4,000,000,000,000 in annual ai infrastructure spending by the end of the decade.

Jon: yeah.

Craig: you know, this projection suggests a massive investment in ai related hardware, software, and other service over the coming years. so, you know, i think governments are afraid to lose the ai war. so they're investing massive amounts of infrastructure in this. the companies, the big comp players are afraid to lose the ai war. the only one who seems not to be afraid to lose the ai war is, apple for some reason. their stuff they announced, you know, several years ago with apple ai still is not working. they've lost.

Jon: they're they're partnering with,

Craig: with gemini. google. google. gemini. yeah. for two years, i think. but they're working on building their stuff in the back end. so we'll see. eventually, they'll come out with something. we'll we'll see, if it eventually strikes home. but if the spend is going 3 to 4,000,000,000,000 and it's a race, you know? we've got a long way to go before, we reach the top. i think we're in a bubble, but based on everything i'm seeing, the bubble is probably four to five years away from bursting. what do you think, john?

Jon: four to five years? that's interesting. so, i lost my complete train of thought here when you said that. oh, nope. i didn't get back. so i think oh, yeah. here and this is how it relates. sorry about that, guys. my concern is that they're integrated now into governments or they're going to be integrated into governments. they're they've got a lot of infrastructure set up. my question is, are they too big to fail now? are they at that point where they're too big to fail? are people so are they too integrated, or will they be too integrated for when the bubble burst that they're gonna be bailed out? what do you think?

Craig: well, that's good to hear. scale salesforce provides another compelling example of ai's practical benefits. their engineering team has become 30% more efficient using ai for coding task. they say the productivity improvement demonstrates how ai tools are delivering measurable returns on investments across various business functions. yeah. but as you drill into stuff like this, there are huge benefits when you start a coding project. it can build the stuff in the back end. but the fur the more complex the code gets and the more you get into it, the less and less efficient it gets. and the more and more you really need to understand what's going on as a coder to be able to solve and fix problems. so i wonder you know, we're just at this point where everyone's saying, yeah. we're getting these big benefits, but that's because, you know, your study is only so long. you know, how long does it reverse to say, you know, debugging ai code takes me 30% more work than debugging human code. therefore, you know, it's going in the other direction. i don't know if it will or not. i'm just saying, i think we get a lot of very positive ai news, but the negative ai news doesn't seem to get the same amount well, hype, obviously. the positive ai news is looking for money. you know, the mit study that says 95% of ai projects, stay in pilot mode. they never come out, and there's a bunch of reasons why they fail. then we hear stuff like this. that's what i mean. i think we hear rosy ends of the spectrum rather than, you know, the true stories that says, okay. we implemented 10 pilot projects. five of them, we went to production. five of them didn't for various reasons, and they're back in whatever. so overall, we're doing this. we get these snippets that say our coders become 30% more efficient. don't get me wrong. i love ai coding tools. at flowgrammar, we do last week, we demoed building an app and publishing within thirty minutes, and everyone got to try it right there. it was great. but like i said, as you get more complicated, these things get more complicated. so i think we're a little ways away from, really hitting that bubble and hitting the limits and what the shakeout in the industry. and when i say bubble, i mean, bunch of these companies will survive. there really is no, going back. but how many of them? right? do we need all of them?

Jon: well, i think we do because they kind of each serve their own niche purpose, especially perplexity, and then tropic and chargebeet unless they can overcome that. and by the way, chargebeet or not chargebeet. anthropic just released opus 4.5, so we'll have to see how that benchmarks against, everything else. but, so i feel like these, these models that can write code, they're like beginner coders right now. and although they're better than beginner coders. but as they get better, i think they will surpass us. and at what point i don't think that we're gonna reach the limits before we achieve, like, a near perfect coding model, if that makes sense.

Craig: it does. i'm just not as optimistic. but i think, you know, at the same time, it's a system. so, you know, when i'm coding, i'm only spinning up one coding agent doing stuff. when i see some of, my fellow compatriots out there who do more of this, they're spinning up ten, twenty, 30 agents to do coding at a time. they're all working with each other, checking each other's code, and and they're they do amazing job. but, you know yeah. the same person, you know, says they blew through their 50,000 pro credits in about three hours. so when you're coding this so yeah. you know? yeah. and they have five to, claude pro $200 max accounts or whatever to run all this stuff. so, you know, there's a difference of what you need to do it, but, eventually, that will come down to a point of everybody will have access to that level of complexity. the thing is the money is not stopping. yeah. like, if the spend is ramping up like this, nvidia sold out, their pipeline is full. and we know that the us and china and even canada, we're building out huge ai data center infrastructures that require all these cpus. it's not stopping, you know, anytime soon. that's why i say we're four to five years away from a bubble bursting. and i think the bubble bursting is gonna happen when there's just not enough investment anymore, and that's when this stuff shakes out. so i right. we got perplexity. we've got claude. we've got, you know, openai. we've got google. we've got whatever microsoft is doing with their stuff. we've got cohere. we've got, facebook's, models and meta. so there's a lot llama. yeah. they're llama models. so there's, again, more and more and more, and they're all spending their models doing these things. that's great. but as a consumer, you know, you go look at the openai models list now. it's way different than it was a year ago. live text, ocr. we do this. we have a code three or four different coding models. that's what i mean over time, they're all gonna be able to do these things. we say, okay. but, yeah, this one has a specific niche with search and research. great. but how long until google can beat them?

Jon: hello?

Craig: yeah. i'm still here.

Jon: oh.

Craig: did you hear that when i said how long till google could beat them?

Jon: oh, sorry. i didn't hear them. i just hear till google can beat. i didn't hear them.

Craig: yeah. you know what i mean by these. you're anthropic is a small player. they just can't spend the money that these other companies can even though they're projecting,

Jon: so i think so i disagree with you. i think if you think about it like this, and maybe i'm wrong, but i think it's more like when you're playing a video game and you choose a character and you build up those characters, all of their, like so each character is different, inherently different. i think that's how ai is going to be at least throughout the bubble and, when that bubble burst. and it talking about the bubble, i don't think it's gonna be as impactful as the comm bubble burst happened. i don't think it's gonna be that bad.

Craig: okay. so when you say something like that, you know, you're probably talking about the number of companies that go under versus other ones. and i look at it and say, it's gonna be

Jon: just that. i'm talking about the impact on wall street too.

Craig: i think the impact on wall street's gonna be 10 times what the, yeah, the comm bubble was because people are spending trillions of dollars on ai, and there's no way they can all return on those investments. so we're gonna see a shakeup. it's just the way it is. yeah. so, you know, we'll have to wait. you know, though, there is worry, like you said, about circular funding, nvidia investing companies like openai who turn around and buy chips from nvidia. critic story that artificially inflates the demand. you know? can big tech really spend 400,000,000,000 a year forever without seeing massive profits immediately? so but they afford to be behind? you know? so the other thing we kinda wanna realize is that even though the numbers were great, the stock price actually dipped slightly 1% after earnings. this shows that investors have priced imperfection. if nvidia isn't perfect, the market gets jittery.

Jon: let's talk about that for a second. so the guy who predicted the two thousand eight crisis, michael burry, he's he's he's shut up. he's closed shop. he's like, there's no way to win in this. peter thiel sold a $100,000,000 of nvidia stake. softbank liquidated $6,000,000,000 in their nvidia position. and 53% of fund managers believe that ai stocks are already in a bubble, which we talked about, which is fine. but that's fund managers who are in the know too or at least more in the know than we are.

Craig: yeah. so i think, you know, there are still large plays in the ai industry, but maybe not nvidia. nvidia is a bellwether, and the market say, you know, there are only a few things, assured in life, death taxes, and nvidia, earnings. so if you think that nvidia earnings are always gonna be up, up, up, they're always priced in. yeah. and here's no way for an investor to get money back is what i mean. it's not yeah. going if it doesn't make it, everything drops. but if it does make it, things still drop.

Jon: yeah. here's the thing that we haven't even touched on too that could affect this is the chinese ais, llms.

Craig: yeah. like deepseek, and there's so many, and there's so many. yeah. and we have a few here in toronto, like cohere. there's grok yeah. canadian with the q. so there's lots of stuff around. yeah. but nvidia's, revenue is up 62 and a half percent year over year, which kinda indicates that it's not going anywhere. so i think the verdict is pretty clear. the money is real. the chips are selling, and ai infrastructure build out is accelerating, not slowing down. yeah. so is this the fourth industrial revolution or just a very expensive experiment? let me know what you think in the comments, and we'll keep watching this story closely.

Jon: yep.