The AI boom has reached a critical turning point.
America's biggest tech companies is pouring money into building out massive AI infrastructure.
So far investors have supported the spending, but history does show that large capital investment waves often lead to too much competition and sometimes weak stock performance.
Now as the AI race pushes big tech toward a more asset-heavy model, some analysts say it may be smarter to look at smaller AI.
With lower costs and perhaps more reasonable valuations.
Well joining me live here at the New York Stock Exchange is Kai Wu, founder and CIO of Sparkline Capital.
Kai, great to have you back.
Thank you so much for joining me.
Great to be back.
Thanks for having me.
Well, here we are, the final trading session of October, and a lot has happened so far in 2025, and this week we saw some key names from big tech companies report earnings.
And this morning we're seeing Amazon as well as Apple higher, but yesterday we saw names like Meta and Microsoft lower.
So when it comes to perspective, where are we in this AI cycle?
Yeah, so I think we did have big tech earnings come through the four biggest spenders which you mentioned reported yesterday and the day before.
And what was interesting is that up until now it's been pretty much a bullish signal.
The more you invest in AI, the more we're going to send your stock price higher.
I think yes, 2 days ago with Meta and Microsoft, we start to see, we started to see a bit of a vibe shift with regards to investors trying to say, hey, you know what, like, You know, meta, you had otherwise pretty good quarter, but you're spending a lot of money on this AI, and we don't know what the ROI is going to be.
Stock down 11%, right?
So I think you know these companies are obviously very committed to spending in AI.
They view it as an existential threat to their business.
The arms race is continuing apace, but you know we're also now starting to see investors again exercise a bit more.
You know, kind of asserting approach towards where the money is actually going and I know that you've done extensive presentations on this subject, but if we do have to compare the current AI build out to other booms, I'm not Warren Buffett, but say railroads or even the internet boom, how would you say this is playing out so far?
Yeah, I mean, it's still obviously early.
I think the benefit of history is we can kind of see the postmortem of how things ended up, right?
So you're referencing the railroad build out in the 1860s, the 1970s.
That was when, you know, post-Civil War we saw a huge expansion of railroad tracks that connect the United States.
33,000 miles of track were laid in just five years, but it didn't end up so well for these companies, the panic of 1973 and then 1983, after the success of. we saw just hundreds of bankruptcies of these real world companies that were just trying to lay all this track where demand was kind of too slow to materialize.
Again, go to the dot com boom.
Obviously we know the internet and railroads changed the world.
This technology was indeed indeed transformative.
The problem was that the companies that actually built this out, the telecoms in the case of the dot-com boom, you know, AT&T, Global Crossing, these companies either went bankrupt, they just didn't make money.
Because what happened was this classic capital cycle phenomenon.
So the capital cycle is when you have an attractive investment opportunity, all these companies, you know, rush to invest and build out the infrastructure to do that.
These are multi-year investments and what ends up happening is that in the enthusiasm, maybe it's rational for each individually do it collectively it ends up becoming a frenzy.
They overbuild.
The demand is too slow to materialize, there ends up getting glut of capacity, right?
95% of fiber was unused.
A.com boom and prices collapse with, you know, bandwidth prices falling 90% and that's not great for these companies.
Maybe it's good for their customers, which is a theme we can talk about later, but for the actual folks who are putting billions of dollars into building out rail or fiber, it's not great when prices go down 90%.
Yeah, so Kai, I do want to ask you about this.
So when big spending backfires and also when we're talking about investing in artificial intelligence, it doesn't have to be these big tech names, right, like Nvidia or Apple.
Because we see these headlines about how these companies, their market cap has topped 4 trillion and possibly 5 trillion.
So moving forward, what sectors within artificial intelligence do you think are better places and why?
Yeah, I think it's very important to to segment the AI stack, so to speak, the value chain, and think about where will value accrue.
Go to the example of the railroads.
We found that all the railroads pretty much went bankrupt like Penn Central and And such, but their customers thrived because now you get access to really cheap rail.
Same in the dot com boom.
We saw Netflix and Facebook grow on the back of the subsidized internet.
So today, I mean, where are we today with AI?
Basically there's Nvidia, the chip supplier, which is the Cisco of today.
There are the hyper scalers, the folks you mentioned building out the data centers plus core weave and a few neo clouds, and then there are the customers, right?
You and I, we use GBT on a day to day basis and we like the fact that it costs. $3 to give us a dollar worth of tokens.
It's a subsidy.
It's great, but that won't last forever.
And so that's the question which is, you know, maybe as an investor you want to think about steering away from some of the folks who are putting the most capital at risk, putting all the money to build out the technology, and think instead of what the folks who could potentially benefit from the adoption of this technology, using it to make their businesses more effective, more efficient, to expand, to cut back on, you know, personnel, things like that.
That's an interesting place to look.
Yeah, and you mentioned chat GPT.
So when you think about the typical consumer out there, they've had exposure to Gen AI, and depending on how big your company is, perhaps you're experimenting with agentic AI, and there are questions about how quantum AI will come into play.
But when we're looking beyond just the end of this year and beyond next year, what are some key innovations as well as trends that you think will actually play out here?
Yeah, I mean, I think I think the biggest thing is just for enterprises being able to actually integrate the technologies into their workflows as consumers, you know, it's low stakes.
We can ask the question, you know, if it gives us a weird answer, we just kind of ignore it and move on with our lives.
For businesses, I mean, doing mission critical things, that's really hard and you know we're going to have to see more progress.
We're going to have to see these tools become better at utilizing.
Other other tools like you know browsing the internet and such, but that's all going to come.
I mean it's only a matter of time before this gets worked out.
I don't know if it's going to happen overnight.
I think that there are some folks in the valley who are over optimistic in terms of the timeline, but I mean you have to take a long term view and say, hey, in.
1020 years.
I mean, the chance that the world is not changed by AI is in my mind pretty low.
Yeah, so before we move on to crypto, one final question regarding AI.
So who do you think the winners will be in this AI boom and what are some of the challenges for the asset heavy companies out there?
Yeah, so I think the the Well, the winners from a stock performance standpoint versus the winners from a technology standpoint are very different.
I think that's important to recognize that we can have a transformative technology, yet the certain group of companies may make no money, you know, as per the historical examples.
I think the challenge with the model developers, um, you know, the the open AIs, Microsoft, Amazons of the world is that um.
There's just so much money being spent on this technology, right?
At one point, these companies, these big tech firms, had this kind of oligopolistic situation where they kind of divvied up the world and had their own, their own fiefdoms.
What's happened now is these folks view AI as a technology that will collapse all the markets into one, where now we're all competing over the same pie, and as a result, you have folks like you know Larry Page at Google saying, you know, I'd rather go bankrupt than.
Race, it's become almost an ego thing.
It's almost an existential risk for these guys and hence we've seen this escalation in capex, right?
It's a 70% year on year increase in capex based on this quarter set to expand to $400 billion a year moving forward, $5 trillion collectively over the next 5 years.
This is a lot of money being spent here and what capital cycle theory teaches us is when a lot of money comes into these sectors, there's a huge.
Of an overbill which would not be good for the folks doing it.
Yeah, and Kai, for viewers out there who are watching this right now and they know that you're entrenched in this research on a regular basis, but they're just saying, hey, how do I actually avoid some of the pitfalls of the AI hype cycle?
What would you say to them?
Yeah, I think you need to, most investors who are invested in the stock market invest via index funds, the S&P 500 for example.
Remember, one third of the S&P 500 is in these seven companies, the mag 7.
That's, you know, definitely a concern given that they're also the most profligate spenders on this big bet, which maybe it works out, but if it doesn't, that's a lot of, you know, a lot riding on that.
So I'd say like trying to tilt away from a concentration standpoint from these 7 companies, you know, on the margins could potentially make your portfolio a bit more robust.
OK, and I do want to get to crypto before I let you go.
So today is October 30th first.
So much for October, depending on where you are in the world right now, you're counting down to November.
So tell us your take on that liquidation event we saw October 10th, 11th, and what this tells you about the current cycle.
Yeah, I mean October 10th was crazy.
It was like an extinction level event for many people.
Um, you know, basically what happened was we had that Trump announcement late on Friday, which in and of itself wasn't a big deal.
It ended up being kind of reversed by the end of the weekend, but the problem was that it triggered a cascading deleveraging of the system.
And in particular to not to get too much into the weeds, but there are these, you know, perpetual futures that a lot of people trade to get leverage into the system.
What ended up happening there was we started to see prices fall and then automatically leveraging where exchanges, they have these rules in place to protect their own capital that will kind of force you out of a position if they start.
Suffering losses and then that happens and it kind of triggers a vicious cycle.
We saw prices collapse.
I mean Bitcoin and Ether are down big, but if you are like me looking at kind of the small cap guys, I mean, most of these things are down 50% to 75% in like a single hour.
I think Adam, a $2 billion do token, was down 99.9%.
It was trading at like a cent on Binance, right?
So I mean you just saw this crazy unwind, but it was all technical, right?
None of this has anything to do with fundamentals.
It, you know, looking back, like obviously we can say that the system had too much leverage in it, that the way that these perfect exchanges do the the auto leveraging that could be fixed and improved, but you know I think I think in terms of the Effect on fundamental long-term investors.
I don't know if it really changes, you know, you know, one's view on the space.
I think you know that it should be pretty much the same once the kind of, you know, wreckage is sorted through and you know, which might take a few more weeks, a few more months, you kind of get back on track.
Yeah, so Kai, finally, since you mentioned the fundamentals of crypto, of course there's a lot that we're paying attention to from a broader market perspective, but in particular for crypto, what matters and are there any particular Price targets that you're watching for any of the crypto majors.
Yeah, I think the big thing to continue to watch is this theme that we talked about last time, this convergence of tradfi, you know, we're sitting on the New York Stock Exchange and crypto.
This is a theme that you know has been, you know, people been talking about for years but really accelerated with the current administration and into the past few months with the ETFs bringing crypto into tray rails and tokenization, you know.
Credit stablecoins, and stocks now bringing you know traditional assets into the crypto world.
I think as that continues to continues to pace and we can get into it, but there's been all these announcements like Western Union folks, you know, in the traditional world partnering with crypto firms to kind of build this out, I think that's a big catalyst for prices in the long term.
OK, Kai, well, it was great having you back here at the New York Stock Exchange.
Thank you so much for breaking down what's happening in AI, as well as your perspective on crypto.
Thank you.
My pleasure.
