Transcription
Welcome everyone. Well, it appears that the two-year joy ride is over because multitudes of data over the past couple of weeks is validating that the trillion dollar US AI project is bust. And today, we're going to break down all the factors that show the US is not only well behind the curve, but have completely gone off the rails.
Yeah, there's an awful lot to discuss in this and I wanted to focus on initially one aspect because then it it actually starts to create uh or actually exposes the reality of what this these AI models are in the United States. And that relates to the fact that increasingly there are US companies and developers who are actually adopting Chinese AI models including the likes of Deep Seek but other also models from other companies and and they're experimenting with it or deploying the Chinese models because they're actually very capable of of doing what the US models are doing. Has there's always been this idea that China's lagging in terms of its AI capabilities uh compared to the United States and that's actually of course being proven to be incorrect.
The other thing of course is it's significantly cheaper than all the proprietary US alternatives and there's even admissions now in the west that the performance gap between US and Chinese AI models they have they sort of gradually go it's narrowed considerably they don't want to ever have to sit there and go well actually maybe China's now leading in some ways so why are you know companies adopting Chinese models and there are a number of factors and of course yeah lower costs is one of them because Chinese models offer substantially lower prices than US models. There's also open-source availability as China said we're going to make the technology available which allows companies to self-host and to customize uh the application for their own needs. And also there's been this big rapid de uh I would call an improvement in in the capabilities of Chinese models. I'd say rapid development which is actually making them in a competitive sense better than their US alternatives. And that you know is with respect to things like coding and and the capability of these models to undertake the tasks that that you know various companies would want to to utilize.
I mean and still of course there are going to be people in the west who say but the US dominates areas of the of the whole AI ecosystem. I mean, and of course, we're increasingly seeing that the United States is incapable of deploying data centers with these grossly exaggerated claims because it doesn't have electrification in process. It doesn't have the the water sources needed for cooling etc. And so yes, we are going through this transition phase where yes, US enterprises do to some extent uh and will do so utilize US companies like Open AI for example. But that sort of perspective of where it was the US dominating in in the United States is changing and and of course the US will be desperate to say oh but you can't trust China's AI models. So because you've got data governance security issues and uh and obviously you need to continue to to stick with the US alternatives. Well, they're clearly seeing that that is not the case. And the argument is at what point does does that sort of become more of a serious problem for for US companies as as opposed to their Chinese counterparts who may increasingly of course uh gain more and more of the business for the reasons I've just stated.
Yeah. And I think one of the most important things we have to take a look at is uh not just the push by the government and this goes back even to the Biden administration when they did the chips act. So they started uh putting in over a trillion dollars towards uh US dominance in the chip industry. This led of course companies like Nvidia uh Intel and stuff to really boost up uh not just the markets but the uh rapid acceleration in chip um innovations. But with that being said, once you have those chips then they pushed in forward into the AI boom. And we know early on uh when President Trump was bringing over some not just sovereign uh leaders like NBS but uh some of the hedge funds from Japan etc etc that were going to allegedly help fund the AI boom. Well, that quickly turned into Wall Street. And we see the market caps of a lot of a lot of these AI companies that haven't even yet gone public, but there's projections of their trillion dollar market cap, etc., etc. But the reality is is when you invest that much into a company, you expect a return on investment ROI. Well, in uh just the last couple days, one chief economist for a money uh money manager came out with a scaling review of the failure of AI to boost profit margins for any business outside of tech. And we're talking, if you remember, there was uh uh companies from restaurants to uh barber shops to everybody was talking about our AI footprint and everything's going to go into AI. Ford, of course, is a great example. They got rid of 200 engineers and replaced them with AI and then they found out the AI couldn't uh provide the the service and they've had to rehire all the engineers. Sorry there had to use the cough button.
Um, but the important thing is is you're there is absolutely no return on investment. And you know another example was is was given earlier today in an interview where the tokenization program that is being used by US models they're selling access to the agents and the tokenization for use of the AI platforms for businesses and these businesses have found that they get absolutely no profit return at all for all the money that they're spending on AI. Now, if uh one of the key things that was said and this is a really um fascinating uh point was uh if the models were truly productivity gains, wouldn't they seek to get an equity or share of the profit from these companies rather than sell access to millions of tokens? So the entire model from the very beginning has been completely screwed up. And that's where you're seeing now businesses and I I know Paul's going to go into this. Businesses are very quickly uh disconnecting from US AI models and they're going over to China.
Yeah. And the big problem is with this is it's many industrial sectors including manufacturing, construction, banking etc. are clearly already seeing as Ken said not really meaningful productive gains or uh from adopting AI. So there's this massive AI spending. It's not producing financial return the way people were told it would do. And obviously we know markets are getting way ahead of the fundamentals of businesses. We've got ridiculous equity valuations now all on the basis of this AI revolution that clearly isn't delivering what it what it was supposed to deliver. And it's not just AI related stocks either because it's spilling over into broader equities. And therefore, you know, the benefits you're supposed to derive from AI are not, of course, surprise surprise, filtering into the real economy. And then, of course, there's also problems with companies who want to adopt AI being able to maximize benefits from it because of integration with their existing businesses. Oh, it costs far more to implement than they understood it to do. And obviously there's also the internal problems where trying to gain accessibility to all the data in corporations and how you might integrate that into an AI model is going to be extremely difficult to to achieve. So you know the the the question therefore is are businesses going to be sat there going well do we throw more you know money good money and inverted commas after bad? they're going to start to already pull the drawbridge up going, "Well, we don't want to spend huge sums of money on AI. We're going to reduce the spending and and and therefore what does that what's the implication of that for this infrastructure that's supposed to support AI?" So, anyone who's massively invested in data centers might start to go, hang on a minute, you know, what's going to happen to semiconductor demand because there's clearly been these projections. we're going to need massive amounts of semiconductors. If all of a sudden you don't need that, then it's going to start to impact the the the infrastructure and the businesses that are supposed to support this AI revolution. So, it's very clear that people were sold an idea and it's not materializing it and it's not actually improving business efficiency. is actually having a detrimental effect as Ken said with regards to Ford and and therefore you know there's going to have to become this realization that well a it's not benefiting businesses a when we compare fundamentals to to what the market's valuing the the AI stocks directly and then what it does in an ancillary capacity with the rest of the economy. it just isn't stacking up and and this therefore is a huge problem because there's no doubt that and it this the Trump administration were probably sold a big idea that this is this is how we're going to save the US economy and it's exactly the same kind of nonsense and I remember it extremely well the dotcom revolution where trillions of capital was was basically eviscerated with very little return on investment. Only certain companies even survived. And and again, it was sold on the basis that this is the future and we're you're going to make massive profits and throw huge sums of money at something. And no one had actually even thought this through. No one had any idea how you were supposed to make money. and businesses got ridiculously high valuations and enormous sums of money given to them via IPOs and they just wasted it on frivolity frankly and vanity projects and and there are not exactly uh parallels per se across the board but there are definitely I'm already seeing similarities between you know the dotcom revolution that failed and here we are with AI you 26 27 years later.
>> Let me give some stats and some examples of why for all intents and purposes the AI race is over. It's done. Uh data came out of a catastrophic collapse of US AI dominance. In one year, American models which share the open router uh system along with China plummeted from 72% uh uh dominion to 33%. Meanwhile, Chinese models exploded to 47%. Then of course we have a very fascinating thing. Uh of all the money that's been put in there, Meta or uh Alphabet, oh I'm sorry, Alphabet's uh Google u Meta has pretty much canled its investment in AI. They're going to be more looking at data centers and trying to sell uh processing power rather than building their own AI. But here, look, take a look at this. Tech companies are increasingly utilizing Chinese AI to build and power their own models. Apple which uses Alibaba's cloud quen uh and BU's Ernie to anchor its uh its AI features. Microsoft which has exploded is you integrating models from deepseek and Z.AI. So when you take a look at uh the the many of the major tech companies in the US are using Chinese models to then try to build their own models and that's going to you know get them so far but they're going to fail because they're not doing any innovation of their own. Uh finally, one of the more fascinating stories that has come out is we have all these chip companies and the superiority of of Nvidia's and TSMC and the like. Well, kind of interesting that Apple now is trying to buy Chinese chips. And I'll go ahead and pass that back to you, Paul, because I think you might have some insight regarding that.
>> Yeah, I mean, look, the the whole basis of this is this. So-called AI revolution has created shortages of memory trip uh chips even and it's caused therefore prices to rise sharply and it's forced companies like Apple who obviously have to because chips cost more they have to raise product prices that's not good excuse me for their business model so Apple's obviously concerned rightly about uh rising costs of chips so it and actually went and lobbied the US government to have permission to source chips from Chinese manufacturers despite the fact these companies are on this Pentagon blacklist because the argument is they have alleged ties to the Chinese military. This was an inevitable consequence of of what's happening and and and obviously memory chips are evolving obviously in in this very sort of bottleneck that that's being created by this AI re revolution because there is massive spending on AI infrastructure. It consumes huge amounts of memory. it does push up cost across the board and then of course that creates inflationary pressure in the process. So this the fact that the Apple has made this request to the US government is because there is a memory chip shortage. So AI has no choice but to look at alternative supply sources otherwise its product base has to massively increase in price. And there's an argument therefore for Apple who then will be looking and going, "But hang on a minute, we're becoming increasingly less competitive compared to, for example, our Chinese counterparts." Because if Apple has to raise prices on things like iPads, etc., then people may start to go, "Well, I'm not going to buy these anymore. I'm going to look at other alternatives. Maybe I'll go and buy Chinese hardware instead." I mean and the argument of course is while there are the shortages memory chips and prices are going to remain elevated. The the argument therefore is in terms of inflationary pressure, how where does to what extent does this permeate the the broader economy and not just the likes of of Apple for example because that does have macroeconomic implications because you're going to get this cost inflation driven by these semiconductor shortages. And where do to what industries do these apply? It's not just smartphones and laptops and PCs. It's industrial equipment, vehicles, anything that's got chipsets in them, which is uh in affects enormous swees of industry is going to be impacted by this. And therefore, you're going to have this in inbuilt inflation that you can do nothing about. I mean, and therefore, of course, because Apple wants to get them from a Chinese supplier, then the Americans start foaming at the mouth about national security issues and oh, we've got supply chain problems and but this is the the the economic reality of having constrained semiconductor supplies. And of course, what's what can we note in this? Chinese memory manufacturers are becoming increasingly having a more relevant stake in global semiconductor production and as shortages persist. Therefore more and more nations are going to go well we need to get them from somewhere. China sat there going well we we now manufacture these and part of this has come about because the US declared war on them with semiconductors. What did China do? They went, "Okay, fine. We'll manufacture them ourselves." And there's a kind of bitter irony that um that now you've got companies like Apple going to the Chinese or wanting to go to the Chinese and and request chips. I mean, talk about the United States shooting itself in the foot. And then another point worth making with regards to the whole thing of AI uh infrastructure, it's obviously having all this investment has knockon effects way beyond the technology sector because you're now getting higher hardware costs. We're now going to get consumer inflation because of the you're going to have supply chain vulnerabilities. And then all of a sudden the the United States thinking it was king of the castle with semiconductor manufacturing and accessibility is going to find itself increasingly in a more vulnerable position. I mean and again there's no lessons learned from the fact work with the Chinese. Don't try and oppose China because in the end it has this awful habit of blowing up in the US's face in doing so.
>> Indeed. Well, I don't have anything else to add. So, uh, unless you got some closing thoughts, I think, uh, this might be a good place to end it.
>> Well, I think just the summary is that, you know, whatever was promised to be delivered is not panning out. Now, okay, someone may argue in a year's time or two years time things may improve. Well, you can never rule that out, but there isn't any real evidence to support that viewpoint. And generally, if you look at all the data and everything that that Ken and I have discussed, and there's a whole raft of other data, there is this collapse in so-called US AI dominance. It's absolutely irrefutable. And and to the contrary uh to this dominance of the United States, Chinese models and their adoption is growing rapidly. And therefore the whole tax strategy coming out of the United States is failing because the United States is rapidly losing ground to China. And we put out some a podcast quite some time ago basically reiterating well sorry just basically projecting that in the future this is exactly what's would happen and here we are it is happening. Okay, things can change and you can never say anything set in stone. But then the question is how does the United States turn this oil tanker around in its favor? And that's another discussion for another day. So with that, uh, thanks very much for listening, for all your ongoing support, which is greatly appreciated. Please like, share, subscribe, hit the notification button because people keep saying they they aren't aware we're putting uh podcasts out. Comment. We really appreciate your feedback and obviously appreciate you helping us to grow the channel. And with that, I'll say goodbye.