So far in 2026 the Mag 7 has fractured and tech valuations are sitting at multi-year lows.
Meanwhile traditional software stocks are plummeting on AI disruption fears and Wall Street is quietly pumping tens of billions of dollars into debt markets to fund the next gen of AI infrastructure, while the defining theme of 2026 so far for the sector is the massive gap between AI hardware winners and software losers.
A SaaS-pocalypse wiping billions off the market caps of legacy software stocks, while joining us here at the New York Stock Exchange to break down the divide between the winners, the losers, as well as the lenders. is Kai Wu, founder and Chief Investment Officer of Sparkline Capital.
Kai, good morning.
Thank you so much for joining us.
It's great to be back.
Well, since the last time you were here, a lot has happened, in particular the conflict in the Middle East.
But when we hone in on what's happening in terms of sector, what are we actually seeing unfolding when it comes to tech?
Yeah, I mean, the markets are pretty interesting this year.
I think towards the end of last year in October, there was a shift in sentiment away from the Magnificent Seven.
As you mentioned, those stocks are down 6% or so year to date, while the Equal Weighted Index is up 2%.
So even though the market's more or less flat, which is actually quite impressive given all the geopolitical stuff going on, we've seen this bifurcation with some of the big boys dragging down the index and the smaller cap names doing okay.
Also the SaaSpocalypse, which you mentioned.
Traditional energy defense stocks are doing quite well, while some of the new economy software names, which were really the winners of the past decade, have struggled.
The software index being down 30 plus percent, guys like ServiceNow, Adobe, Salesforce down even more.
So it's quite an interesting bifurcation and a shift in sentiment from the uniformly bullish bull run we had over the past few years.
Yeah, and sas-pocalypse is a new word for our lexicon here.
So do you think some of that sell-off is overblown right now?
Absolutely.
I mean, I think, well...
So part of the sell-off is just a renormalization of expectations from an era where valuations were probably a little bit elevated, right?
The business model of software was considered kind of the perfect business model all investors wanted in.
That sort of lost its luster.
So some of it can just be seen as a reset, say of kind of like, historical expectations.
That being said, I mean, the sell-off is very violent, as I mentioned.
Some of these names down so big, Ford P multiples, or 1510 in some cases.
We've even seen some single-digit P ratios.
And so you have to ask the question, which is, you know, what is driving the sell-off?
What appears to be driving the sell-off is this fear that AI, through its, the ability to now vibe code software, and to 10x the productivity of software engineers, will compete away the most that these companies have.
The question I would ask is, was the moat of a company like Salesforce ever its code?
Or was it perhaps instead the network effects, the brand, the switching costs, the kind of embeddedness in the workflows of enterprise Fortune 500 companies? that that is actually more the case, that yes, these companies will need to make shifts to how they price, and obviously there will be an adjustment period, and potentially the overall business is a little bit less good than it used to be, with higher marginal costs for AI than pure traditional software.
That being said, these companies I don't think are going away, and just given how dramatic the sell-off has been, there almost certainly are opportunities to kind of bottom-fit for some of the winners.
Like yes, some of these names are likely challenged and may like the blockbusters of the world face significant issues but there are others that are almost certainly diamonds in the rough that are being sold off despite actually having pretty solid businesses where they just need to make a few adjustments to adapt to some of the changes we've seen.
Yeah and Kai as you mentioned there have been a lot of comparisons being drawn to what we're seeing unfold in the AI revolution to, say, transportation, railroads, and the advent of the internet.
So you mentioned the word opportunity.
How are you separating the winners from the losers?
Yeah, so I think you could even step back, and this goes back to what we discussed last time, which is, if you think about the whole AI value chain, like where is the bottom all the way to the top, right, on the bottom you have, I guess, energy and infrastructure, and you have chips, so like NVIDIA and those guys.
Then you have the hyperscalers building out the data centers, plus CoreWeave and some of the NeoClouds.
Then you have what you would call like the model layer and then the application layer.
So the application layer is what's kind of the most interesting right now.
Up until October last year, all the energy was around the build out.
These companies announcing bigger and bigger plans to roll out increasing CapEx, building out more and more data centers, trillions of dollars being spent there, which of course buoyed stocks like Nvidia, some of the hyperscalers and utilities and energy companies.
Now the big question, starting entering this year, and we're approaching the earning season, we'll see how this plays out, is to what extent are companies actually driving ROI from their investments in AI?
So it's great that you're spending a trillion dollars doing this, but are you actually expecting to make sustainable revenue from this?
And so that's the big question, and that goes to the question of the application layer.
Anthropic actually put up some really put up some really good positive numbers to the extent that they're now Able to get more and more traction on the enterprise side selling, you know tokens to companies trying to use agents So that's that's really encouraging.
The question is what about the traditional companies like the sales forces of the world, right?
Who are now their stocks have fallen so much because investors are saying well Anthropic's gonna eat your lunch because why do we need you now?
These companies aren't sitting idle either.
They're also investing in trying to position, or they have been doing this too for the past few years to be fair, towards more AI-driven workloads.
And so that's the big question.
Which companies are actually kind of sitting around the corner, seeing the potential of AI, and could potentially implement a lot of these things on top of an existing distribution and client base, as opposed to which are the ones which are kind of stuck with innovators' dilemmas or bad business models, who are going to have a hard time adapting to kind of a new world, a new paradigm shift in technology.
Yeah, and Kai, while I have you here, you mentioned Anthropic and last week Mythos was in the headlines and whenever you hear Secretary Besson as well as Jay Powell meeting with CEOs of banks, then that really piques your interest.
So how concerned are you about Mythos?
And we all know that OpenAI as well as other organizations that are in this space are going to be releasing powerful models as well.
But when we think about who the target market is and the limitations of access to models such as AI, Mythos, then what do you expect to see?
Do you think the competition is going to be limited here?
So I think, first of all, the decision by Anthropic to hold back Mythos around fears of security and such, that's kind of like a marketing playbook that these guys have been doing now for years, since like GBD2, right?
Or 3, I guess.
Where they would hold back and say, hey, this is just a powerful model.
And of course, everyone gets interested, and then we only work with the biggest banks.
Which makes sense, right?
Anthropic, as we know, is compute-constrained, especially relative to open AI.
So they can't, I don't think they can really do a full general rollout.
So they're saying, let's prioritize our biggest and best customers first, give them exclusive access.
And so it kind of plays into both their desires, A, for the marketing bump of saying we have this powerful model that no one else has, and also to be conservative with how they're using compute and allocating only to potentially very large customers.
Very smart strategy on their part, but again, it's something that we've seen before.
So Ben Thompson, who's a well-known technology analyst, he talks about it as like the boy who cried wolf, right?
So up until now, people have always been crying wolf, but Ben's point is that eventually the wolf does come.
Right, so, you know, it's an interesting dynamic in what they're doing, and I think it serves, again, multiple points in addition to kind of the security concerns, which I think are somewhat legitimate.
There's the other two points, the marketing piece, and then also trying to conserve compute.
And one other thing I do want to ask you about is earnings season is underway.
Goldman Sachs reported and the big banks will be in the spotlight first.
But as we get to some of the Mag 7 names as well as those AI names that we're watching what are the metrics that you're watching for and what are the key words from the guidance and earnings calls that you're listening to.
Yeah, so the last research piece I put out in January, what I did was I used LLMs actually to parse all the different earnings calls coming up each quarter and saying, what are these companies doing with regards to AI?
Are they implementing AI at all?
And if so, how are they seeing ROI or not?
How precise can they be? report, are they saying we have a 17% increase or a 12% decrease in costs?
So I think what you want to do, and you can use tools for this, obviously, is to be parsing what these companies are saying with regards to AI adoption.
I think the other big theme is this rotation away from Mag7, which we discussed.
Is that sustainable?
We've seen the Mag7 stocks fall on AI capital concerns, competition risk.
And conversely, we've seen the Equal Weight Index do quite well this year.
So the question is, is that all the small-cap tech companies actually seeing meaningful bumps from being AI customers?
And do we expect the rotation away from MagSeven to their customers to be sustained moving forward?
And I know, Kai, that you spend a lot of time looking through some of these data points here.
So what are some areas that key investors should be paying attention to that they are not?
Oh, well, I mean, again, I think AI adoption is very important, but I guess investors are, to be fair, paying attention to that.
Obviously, what's going on in Iran is tremendously important for markets.
I don't know to the extent to which any investor, aside from a few handful of people who may be close to the administration have any kind of edge when it comes to that.
But I do think that all our views do need to be somewhat moderated with this understanding that uncertainty is very high due to geopolitical risks.
And so maybe the amount of risk that one wants to take should be a little bit like, let me pull back a little bit, and also focus more on the kind of bottoms-up areas where you as an investor may have expertise.
So in my case, technology and software is really interesting, AI adoption across the entire economy, not just software stocks. industrials and energy etc. and kind of focus on that what is like the company level story and trying to get away a little bit from the macro which is of course you know very challenging right now and I think for most investors probably too hard to predict.
And Kai finally before I let you go I do want to leave this on a positive note because we continue to hear about all the downsides of artificial intelligence and the risk management and the guardrails that are necessary but when it comes to the innovation part Where do you see us, not just a year from now, but beyond?
Yeah, no, I'm actually tremendously bullish on AI.
I don't necessarily think that the value will true to the companies that investors currently assume it will.
That being said, as a tool, as a technology, you go back through the history of the internet era, you mentioned the railroads, all these technologies, electricity, It's been tremendously helpful for the economy, for the users, for consumers, individuals like you or I, able to turbocharge our productivity on a day-to-day basis, or even plan fun trips or cook fun recipes.
So I think AI actually has a potential to be really transformative, and I can tell you personally, In terms of my own work, use case, I mean, I'm using AI coding agents all the time, and they're really, really powerful.
They give me the ability to have junior analysts at a small shop to have access to a lot of what would normally be expensive costs for a company.
So I think increased productivity for companies, as you discover more and more use cases, you would almost assume that that actually would have a multiplicative effect on the wealth of society and ultimately usher in abundance.
Well, Kai, a lot to keep our eyes on.
So as always, great to have you on the show.
Thank you so much for joining me, and thank you so much for sharing all of your insights.
Thanks for having me on again.
Thank you.
