Staying Calm Amid the AI Panic: Navigating the Hype

Why is it timely?

Well, you've probably seen some of the news out there regarding some major developments on the ai front with Deep Seek, a less than 1-year-old AI company out of China. And it's injecting a fresh amount of turbulence not only into the stock market, but into a number of key companies,

nvidia, for example, that are thought to be the big beneficiaries of A. I. Spending. And we've had some big A. I. Spending announcements recently with President Trump introducing Stargate. We've also learned that Meta is planning a significant increase in its capital spending in [00:01:00] 2025 compared to 2024. So Lindsey and I are extremely thrilled about that.

I'm thrilled to share this conversation because it brings some much needed context, perspective, I think when you're done listening to it, you're going to realize that, you know, disruption is a good thing, but on the other hand, that doesn't mean that the folks that were positioned to win. Aren't going to. So I would encourage you to sit back, listen to our conversation with the folks at SuRo Capital and stick around for what we share with you when the conversation's over.

Hey folks, it is January 27th, and there is a lot going on in the AI space, and I'm really excited, as is Lindsey, to have a special edition of the What Does It Mean podcast. Today, we're going to be joined by Mark Klein, the chairman and CEO of SuRo Capital, trades under the ticker SSSS. It's a publicly traded growth stage venture capital [00:02:00] firm that invests in a wide array of areas, but has key positions in AI, consumer and software investments. Also joining us from SuRo is Evan Schlossman, one of the folks on the team that focuses on AI. Mark, Evan, thank you for joining us today. As I alluded to, it's a big day, given what's going on, a lot of disruptive talk on the AI front with DeepSeek. And I just wanted to, start our conversation off from, gathering your perspective on what is DeepSeek?

Is it the disruption that some people are talking about? How are you communicating your thoughts on it to your shareholders? In other words, help the audience understand what's going on.

Evan Schlossman: So, DeepSeek, it's interesting, it's based on the version 3 of the model, that was released back in December and then on last week they released, DeepSeek R1, [00:03:00] which really takes an interesting approach into how they are training the model. It's taking a more reinforcement based approach which has allowed them, based on

the varying reports to basically train and produce a model that is close to what OpenAI's one model is getting in terms of output.

Maybe not quite there in terms of robustness, but close at a significantly reduced cost, at least in the training side. So very interesting in how they're taking a new approach that appears to be relatively efficient into the training side.

Lindsey Bell: Evan, I just had a quick question on, when you read all the reports, and I'm no AI expert. I'm not in the weeds of it like you are. For me, I see two things, though, as an investment minded person. There's the reduced cost to create the AI, but there's also a quicker speed to market.

Which is the bigger concern or both concerns for the hyperscalers? What's your [00:04:00] take?

Evan Schlossman: Yeah, so I think it's interesting when you look at both, and I don't know if you fully know this speed to market, but the cost and the speed. What we're seeing is, you know, first of all, there's just varying reports. And so, you know, if we take for a moment for granted which there's some ambiguity to it, right?

That they are getting the efficiency in terms of the cost reductions, which market paper reports, you're seeing a change in the efficiency curve. And so what you could see is that for, you know, X unit of input, right? Token, whatever you're spending right now. You're getting Y out. And so with this new model, if you're able to put in, you know, 100 apps, and previously you were getting 100 Y out, now you're getting 200 Y out.

I think it's not clear, one, how this handles with scaling resources. So if instead of, you know, 100 of the chips that they were using, the Hopper chips, or based on that architecture, they're using the Blackwall chips, or if they [00:05:00] double the amount of input, right? What happens? Does it scale? Is it going to increase the output capacity, or is it reaching a plateau?

We don't quite know that yet. But what we're not seeing is a covalence that, you know, adding more resources is not going to produce a greater output. And so if we think about an efficiency curve that just shifted up if you're a hyperscaler and you were investing 80 billion dollars a week ago, you know, if you're investing 80 billion dollars today, it's possible that given a new approach, now you see two, three times the output which would be quite frankly, fantastic in terms of an ROI or use case, as well as just broadening the aperture for companies that now can justify investing more in AI and using the GPU resources.

Chris Versace: Evan, let me follow up on that, because the immediate reaction that we're seeing on these developments is that, some of the poster children for AI, Nvidia, for example, and others [00:06:00] on the hardware chip side are really taking it on the chin. But, at the same time, we're seeing other companies out there, software companies in particular, that could benefit from perhaps a quicker pace of adoption of AI really start to rise on this news.

Does that kind of support what you're thinking?

Evan Schlossman: So I think absolutely from, if you look at the SaaS companies that are doing well this morning after this news are presumably on the idea, right? They're going to be able to adopt this technology better. The ROI in business cases may be more attractive. I think that supports an idea that you're broadening the aperture of companies that are able to invest in this.

When you look at the chipset and the GPU structure, I think Mark's probably talked about this as well. But if you look at recent announcements that have come out in terms of capex spending, it's not clear that we should expect to see a slowdown from any means from these hyperscalers in terms of capex.

So whether or not, you know, the specific chip [00:07:00] from NVIDIA, if the next model is impacted by this type of training structure or the architecture, I think that is something we'll see as more and more evidence and information comes out about the new model. But from the underlying capex spend I think we've seen some pretty good indication of the last week that it's slated to continue increasing.

Chris Versace: So I take it with that, Evan, you're referring to the Stargate announcement. Meta talking about a significant increase in CapEx for 2025 compared to 2024. And I actually agree with you on this. I'm actually looking forward to the comments that we'll get this week and next from other big tech companies, Microsoft, Amazon and others out there just to see if this AI arms race on spending is continuing to accelerate.

Lindsey Bell: I would just add on to that. My question is ROI. What is a good ROI on these investments then? What was it before? And what do you think it goes to now?

Evan Schlossman: I think we think about [00:08:00] this in two ways. One is current. And so if you look at announcements and earnings calls from some of the hyperscalers recently, you are seeing more and more of a focus and more tangible examples of companies that are able to use these AI and LLM models in order to generate real cost savings or value, right?

So if you think about the way. That a company could structure an algorithm internally in order to get value out, especially for advertisers, right? You're seeing value come in from the revenue side, but you're also seeing more and more examples of cost reduction. Then on the flip side of this, and I think this goes back to where we talk about from spending and maybe just to shift the efficiency curve, is you're essentially seeing kind of an arms race in, so to speak, you want to spend to have the best model because the future possibility of this technology is so significant.

And so it's not clear right now that companies [00:09:00] are spending to get to a specific milestone. You know, they're not looking to run a hundred yard dash and stop at a hundred yards. They are looking to see how quickly can they accelerate to whatever kind of the end conclusion might be, but it's likely not a short term end conclusion.

And so even if you have a change in what the models are outputting, that's fantastic. You have a better use case, better ROI standpoint. But if you're spending in order to have the best technology to be a market leader and to not get left behind, that's not necessarily going to shift even if kind of the underlying architecture is improved or updates or there's new version of that.

Chris Versace: Let me step back a little bit and bring Mark into the conversation. I had been familiar with zero capital in the past, with a number of the portfolio companies, but I believe, Mark, that over the last several quarters, you guys have really shifted around your portfolio. You've made several investments in a I, including OpenAI, CoreWeave, [00:10:00] and I believe one or two others.

If you could give us the perspective you had what led to this shift? And do you see more of it happening in terms of the portfolio? In other words, are you going to continue to lean into AI and other disruptive technologies?

Mark Klein: Thank you. before I get to that, the one, I guess I'll reaffirm your commentary about Meta announcing an increase in their spend where Microsoft is and the Stargate announcement earlier last week. Given that DeepSeek been out and around since end of December, it's not a real surprise to all these huge companies of what was going on and they have made this increased commitment to expenditures in light of that, so it's news to maybe a broader audience, but not to those that do, if you will. And the idea of the president, second day in office, standing in front of everybody with [00:11:00] Oracle and SoftBank and open AI and say, there's a 500 billion commitment that open AI is going to run that whole process, if you will, to keep everything US based, et cetera.

That's a massive commitment on top of all the others. This seems like a surprise in the sense that, oh, wow, look what's out there, but all the folks that were planning to spend money. Knew this was out there, still planning to spend money. So the end result of this will be what the end result was, but it is an out of left field that people woke up and said, Oh, wow, look what exists.

Chris Versace: Mark, not that we're conspiracy theorists or anything like that, but it is it possible that ahead of a February 1st conversation about tariffs, that this could be a little saber rattling just given the timing of it all, and given what you said about, the Stargate announcement last week.

Mark Klein: I'll flip it the other way and say that it was an important enough announcement, and it was, it's so meaningful size and [00:12:00] scale and importance that to make that announcement as big as it was at the time that it was on every major news channel. It was, it's just an important announcement to be made. And it really speaks to the importance of AI infrastructure.

And that'll bring to your question. So I just, I wanted to meet that. As we've been around since 2012. Public in 2012. So a lot has changed. We have owned every name that you figure we would've owned over the years. Facebook, Twitter, Dropbox, Spotify, Snap, Lyft, Palantir, Coursera, a litany of really interesting names.

And we enjoyed a great run up as did the market through 20 to 2021. We decided at that point in time, the market was a bit extended and we started to liquidate a vast majority of our portfolio and did so. And we ended up paying out about 250 million of dividends to our investors. And then we sat still [00:13:00] and we came to the conclusion that the disconnect between the public markets and private markets.

It was pretty severe. Private markets had not nearly come down as far as the public markets. And it was a time to just take a breath and see where the puck was heading, if you will, to make a determination, how best to determine to deploy capital. We deployed a little bit of capital in 2023.

And then in 2024, we made a couple of decisions. I would say the most amazing, been around quite a long time, phenomenon that I saw was the supply demand imbalance or need for compute and the availability of compute. And almost, I've been trying to rack my brain to see where I've ever seen something that was, this can't be true because it's so obviously disconnected. Doing all our work, we came to the conclusion that others did, that there's going to be massive spends and massive amounts of money in AI in the first place [00:14:00] you go is picks and shovels and the infrastructure. And that's what we did. So we identified CoreWeave. That was, we actually identified all of them about the same time, but to actually access the social period of time.

You know, accessing private security is challenging, accessing the direct price but we identified Corweave and said, given where they are, where their contracts were, where their customers were, what the demand was, and the growth rate where they were coming from, it would, it seemed like there would have to be something that would massively

change in order for that not to work. And that's why Corweave is probably the most anticipated IPO of 2025. But we invested in it. Even the analysts that cover us didn't know who Corweave was. And So, Corweave really, really exciting. The growth curve is, it's one thing when you see a growth curve, charts that show these massive [00:15:00] growth curves by projections.

And then there's others when they're actually contractual projections. And so that's what we saw at Corweave. That one very excited about committee. That's the largest investment we've ever made in the 13 years as a fund as one was one company. OpenAI I think speaks for itself. Clearly market leader, continues to be the market leader is the reason when you see Stargate and they say, and open AI is managing the whole thing.

Open AI is pretty much a must own if you want to be in the infrastructure space and in AI in general and then in Vast probably name that not to people any people know but is it is wait, you will know. The data is access managed. Utilized for training AI it is better that what's out there by a lot. They have core relationships with NVIDIA and others.

And if we come back in about [00:16:00] six months from now, say, oh, wow, thank you for telling us about Vast. So we went and we committed about a third of our portfolio. Into these names. Really super excited about those. And as you mentioned, we have a couple of consumer names. I'll give you one that, it took all of us a bit of time to get our arms out, but super exciting is Liquid Death.

Chris Versace: I knew it. Mark. I saw it. I saw it in the deck. I knew you were going to bring it up. And I got to tell you, it's a great name and I enjoy the product,

Mark Klein: Excellent. We do too. And it was about the longest debate we've had as an investment committee to decide to do it. Because if you just say it's water in a can, it's water in a can. Now it's not just water in a can, now it's flavored, it's sparkling flavored, sparkling teas and now healthy sodas and all that.

But it's a brand and their branding is amazing. Their growth is [00:17:00] incredible. They did over 300 million in sales last year. They're growing super, super fast. They had about. I guess about a billion and a half dollar valuation a year ago, and it'll be a public company in the next 12 to 15 months. So very exciting.

People knew the name. We're excited that we owned it. We, that was one of our names we added recently. And finally, this is Canva. So no one on Wall Street knows what Canva is because they all know what PowerPoint is. But everybody else in the world knows what Canva is, and doesn't really know what PowerPoint is.

So Canva has really benefited from using AI tools in their product and continues to do so. They now are doing two and a half billion dollars in annualized sales. They have 220 million active users. They recently did a round of financing at 32 billion. That was up from a prior secondary round, 20 percent lower than that.

I think that's another [00:18:00] company that is setting up to go public. Whether it's end of this year, beginning of next year, they're very communicative about the finance. So yeah we went into this year with probably close to 50 percent of our portfolio in cash because we wanted to find the right? places, right time. We felt thematically AI specifically, the the infrastructure and the base side of AI as opposed to some of the applications made the most sense for us right now. Committed a lot of money in that. And then we have a couple of the consumer names. The other name we had, we no longer had was we did own Oklo which was Sam Altman company.

Small module nuclear reactor company. We made 20 times our money. We sold it. And we don't really own public companies, but we had that from a private investment that went public, but that also is a clearly benefiting from the energy, not only the [00:19:00] existing energy needs to drive data centers, but the perceived ongoing energy needs and how that's going to be solved.

Chris Versace: Yeah, no we've been looking at that over at the street pro portfolio on the nuclear side. And I think you're right. We're also seeing other investments. I think Blackstone talked about acquiring some capacity. Microsoft is making some deals and, I think it's fire island in New York is starting to move closer to refire. I agree with you that just the given the explosion and data consumption, data creation that we're gonna have as a result of AI, the power requirements, this entire digital infrastructure is going to benefit. It's not just AI chips. I think network infrastructure is going to have to be upgraded and improved. And just all down the line. So I, I think where your positioned is extremely smart. I guess my question to you, Mark, is you mentioned a couple of eventual IPOs. Just given how you guys work, investing [00:20:00] in private companies, you do need to monetize at a certain point. But does it have to be IPOs or are you guys just as happy with M& A activity taking some of your positions out?

Mark Klein: That's a great question. Private markets are a complicated. There's urban opacity in pricing and just the ease of entry and exit is challenging. It just is. Over the years we've exited in the following ways. We have had obviously IPOs, there's been M& A activity in and around a fair amount of our portfolio.

We've sold back into the private markets, so not only bought through the private markets, sold back into the private markets. Obviously during the SPAC craze we probably have four or five companies go public through that as well. So we can, we have a lot of different ways to exit our portfolio. I think the private market activity has grown quite a bit.

So the liquidity in those markets, [00:21:00] especially bigger names is pretty strong. If we wanted to sell some of the names we just recently talked about, there's a private market. It's for all of them. If we chose to do that we just bought. So the last thing we're really interested in is turning around and selling to the private markets, but IPO is clearly a one and the M&A side will take care of itself through those companies.

We won't have any real influence in driving M&A activity, but There will be M&A activity in a bunch of the spaces that we're in. So what'd be surprised if some of our companies end up bought rather than going public either through a traditional or non traditional route.

Lindsey Bell: So do you have a target or average turnover in your portfolios then? It seems like you're really patient with the opportunities that you invest in. You do your due diligence and kind of wait it out. Do you have an average turnover?

Mark Klein: Good question. First of all, we are clearly, we're [00:22:00] clearly rifle shops, not shotgun approach. We spend a fair amount of time determining what we want to buy. And the later stage investments. We assume that our turnover is somewhere between 18 and 36 months would be on the outside. So the pre IPO names, if you will most of the names, if we buy them pre IPO, and we didn't talk about service Titan, we did own service Titan.

We knew that was going to be an IPO sometime last year. When we bought it we were right. The market liked it, which is great. This is the first one out of the box. But that almost hit our timeline within a few months. For almost any company that IPOs there's a lockup period that we have after the IPO. Unless it's a direct listing, the lockup periods typically six months.

After it was public. So on the later stage, 18 months to 36 months. Some of the earlier stage can be a bit longer, but we don't have a timeline that's five to [00:23:00] seven years out. We're much closer to late mid later pre IPO investing and really early seed five, seven years.

Hope that works.

Chris Versace: You know what I like about this conversation we're having as I think about your investment style, the areas that you're focused in on they're topical, but it also means that, just like most investors, you guys really have to have your finger on the pulse of what's happening. And I think that as we go forward with our podcast, talking to you guys from time to time would be very insightful, not only to us, but really to the audience that we have and other investors out there. So I want to say to both of you, Mark and Evan, thank you so much for coming on and really helping give us some, much needed perspective on what's happening on the AI front today. Very timely, very helpful. But I also think though, too, that the opportunity with, SuRo capital and its shares for investors that are looking for ways to participate in what's coming, I think that is as [00:24:00] interesting. And with that, I would just ask Mark, where can folks go to learn more about SuRo capital?

Mark Klein: Send an email through our IR site. I believe a couple of things. Our jobs was invest the money well, value our portfolio accurately. So, you know how to value it when you buy it and communicate our story clearly. So I've gotten on the phone with small investors, institutional investors. I'm happy to take phone calls to discuss what we're doing to help people understand our story.

We're proud of our story. We are really the way to participate in the private markets with a public vehicle that has public liquidity. That is reasonably priced that really we spend time and effort to communicate it, analyst coverage, and our valuation process is extremely robust.

And we're happy to [00:25:00] tell the story, if people want to learn more about it, of course, we have a website. Go to the website if they want. Really communicative. Our earnings calls we do them four times a year. Unlike some other closing close down funds don't have to where we're BDC report quarterly.

We have extensive earnings calls. We do have analyst coverage. Four firms cover us. So there's plenty of research on us, but if they want it from us directly, they should feel free to reach out to us.

Chris Versace: Awesome. Awesome. We look forward to having you guys back in the next couple months, even if it's only to talk about the eventual core weave IPO. But I suspect it'll be another great conversation. So thank you very much, gentlemen.  

Chris Versace: All right, folks, that I think was a great conversation with Mark and Evan. Lindsey, what did you think? Mm hmm.

Lindsey Bell: I think it was excellent. It certainly calmed the fears that I had about rising competition and uncertainty in the AI space and how to invest in it. But I also thought it was really revealing and a reminder to [00:26:00] investors that the investments that are being made in CapEx, It's you can't get hung up on the ROI.

And I know those are dangerous words, right?

Chris Versace: Mm hmm.

Lindsey Bell: you have to spend to create efficiency and you have to spend to be a part of the game because this is an arms race. And so I don't know, it was just a refreshing or interesting, differentiated perspective. What did you think?

Chris Versace: Well, you know, to me, the point that Mark made was that this has been known to people who have been really focused on the industry people who are deep in the weeds. It's only when it makes mainstream headlines, let's call it, does it kind of percolate across Wall Street's desk, right? And, you know, it just reminds me of times that, you know, you and I see certain things in data. We have a pretty good idea of what's coming, but it's only when I Everybody else sees it, right? That you get the reaction, right? The big mainstream reaction. And sometimes [00:27:00] that's appropriate. Sometimes it's not.

And I think if we play this out with what we're seeing regarding Deep Seek, maybe it, maybe it is overdone.

Lindsey Bell: Yeah, I think that's the name of the game with investing, right? It's finding these opportunities. Learning as much as you can ahead of time, but also taking a deep breath when big reactions like this happen and, you know, taking the time to realize is this a buying opportunity or is this something worse?

And so I think it's, that's what's investing is all about.

Chris Versace: I agree. It's also looking for confirmation, and I think we're going to have a lot of confirmation in the next two weeks as the rest of big tech reports. That's going to be Microsoft. That will be Meta, probably fleshing out a little bit more on their capital spending plans, but also Alphabet, Amazon, and others.

So it's going to be a big set of weeks, folks. I would encourage you to come back and listen to Lindsey and I break it all down. But if you need Lindsey or myself between now and the next, what does it mean podcast march on over to LinkedIn for Lindsey's comments, and you can always find [00:28:00] me at the street pro thanks for listening.