Horizon Kinetics - Earnings Call - Q3 2025

Executive Summary

• Q3 2025 total revenue was $17.9M, up 37% year over year, while EPS was $0.39; advisor-only operating income rose to $5.5M (up $4.0M y/y), reflecting core fee growth amid consolidated investment product volatility.
• Sequentially, revenue declined vs Q2 ($19.8M → $17.9M) as unrealized investment losses tied to TPL’s ~12% price decline weighed on other income lines, though operating margin improved to 16.5% from 12.7%.
• The Board raised the quarterly dividend to $0.106 per share (49% increase vs prior quarter), with management reiterating a ~70% payout of net income as a capital return priority.
• AUM was $10.4B at 9/30/25 (vs $10.5B at 6/30/25 and $10.8B at 3/31/25), with fee revenue supported by higher 2024 AUM levels (TPL, GBTC) and incremental net inflows; consolidated “noise” remains from investment products.
• No formal top-line/margin guidance; near-term narrative centers on dividend cadence, advisor-only profitability, TPL/digital asset marks, and an announced plan to launch a new ETF in Q1 2026.

What Went Well and What Went Wrong

What Went Well

• Advisor-only operating income reached $5.5M, up $4.0M y/y, underscoring core fee-based profitability despite investment mark-to-market volatility.
• Dividend raised to $0.106 per share, a 49% increase q/q; “we are taking 70% of our net income, and we are paying it out to shareholders” clarifies capital return policy.
• Operating margin improved sequentially (16.5% in Q3 vs 12.7% in Q2) as fee revenue growth outpaced expense growth; management emphasized focusing on the operating company’s YTD operating income of ~$16.1M as the key metric.

What Went Wrong

• Unrealized investment losses of $7.0M (primarily due to TPL’s ~12% decline) and equity losses of $2.0M created bottom-line volatility; net income attributable to HKHC fell y/y ($7.2M in Q3 2025 vs $19.6M in Q3 2024).
• Sequential total revenue declined ($17.9M vs $19.8M), as advisor-only revenues softened and investment products’ net losses reduced consolidated other income.
• Prior quarter included a $0.9M non-cash goodwill impairment in consumer products and elevated unrealized losses (TPL down ~20%), illustrating sensitivity to investment marks and consolidated vehicles.

Transcript

Mark Herndon (CFO)

Good afternoon, everyone. Thank you for joining us on this call. My name is Mark Herndon, Chief Financial Officer of Horizon Kinetics. We are pleased to have you join us for our call, and we'll cover our results for the third quarter of 2025. First, a reminder that today's presentation may include forward-looking statements. Reliance on forward-looking statements involves certain risks and uncertainties, including but not limited to uncertainty about the future of security valuations or our performance. During the course of today's call, words such as expect, anticipate, believe, and intend may be used in our discussion of our goals or events in the future. Management cannot provide any assurances that future results will be described in our forward-looking statements. Furthermore, the statements made on this call apply only as of today.

The information on this call should not be construed to be a recommendation to purchase or sell any security, particularly security or investment fund. The opinions referenced on this call today are not intended to be a forecast of future events or a guarantee of future results. It should not be assumed that any of the security transactions referenced today have been or will prove to be profitable, or that future investment decisions will be profitable or will equal or exceed past performance of the investments. We encourage you to read our filings with the SEC on our Form 10-K, as well as our other filings, which describe the risks and uncertainties associated with managing our business. The company does not assume any obligation to update any forward-looking statements made today.

These filings can also be found at the OTC Markets website, and our press releases or other information is at our corporate website at www.hkholdingco.com. Today's discussion will be led by Murray Stahl, Horizon Kinetics Chairman and Chief Executive Officer. I will also be available to answer applicable questions and will moderate the questions. If you would like to ask a question, you will need to be logged in and connected to the GoToMeeting platform. Those of you on a telephone connection will be in listen-only mode. Again, for those of you that are on the GoToMeeting platform, you can submit the question via the chat function. Please direct those questions to the presenters, while I will summarize and relay as best I can so that we can address as many questions as possible. With that, I'll turn it over to Murray to start us off.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Okay. Thank you, Mark, and thank you all for joining us. I will lay out the format for today. What I'm going to do is in a minute or two, I'm going to turn it over to Mark Herndon. He's going to review the financials, and I'll take it back. I might go into a little detail of a financial high point that I think merits note. I will talk about strategy and how we're going forward and some interesting things that we are in the process of doing and some other things that we hope to be doing. With that, Mark, you can describe the financial results, and you can relay it to me after that.

Mark Herndon (CFO)

Great. As a reminder for everyone, our Form 10-Q update continues to be our required GAAP presentation that includes certain proprietary funds as consolidated entities. Our press release continues to present both that GAAP presentation as well as a supplement that provides our financial statements excluding those funds. We call that the advisor-only presentation. Consistent with what we have previously reported, this is a presentational matter that does not impact the company's earnings available to HKHC shareholders or the shareholders' equity of HKHC. The consolidated results present higher total assets as we include the investment assets of those funds that we have consolidated on our balance sheet, as well as a line item called redeemable non-controlling interest. That line item essentially represents our clients' account balances that are supported by the assets of those funds, which you'll see identified in our financial statements as consolidated investment products.

Another notable difference is the treatment of management fees charged to those consolidated investment products. Under GAAP, those revenues are eliminated for consolidation since the fund is presented within the financial statements. It is akin to an intercompany transaction. However, the economic benefit resulting to HKHC shareholders remains, and that economic benefit is reflected through a smaller allocation of the investment returns of the consolidated investment products to the redeemable non-controlling interest than what they would otherwise have received. You can also see the impact of those items in a table within the ND&A of our 10-Q filing. Our results this quarter continue to be favorable for HKHC shareholders. The company recorded revenues of $17.9 million for the quarter, a $4.8 million, or 37% increase from the third quarter of 2024.

The year-to-date period was similarly higher, with a $55.8 million of revenues thus far in 2025, which was 49% higher than 2024's year-to-date period. These increases are primarily the result of the overall increase in AUM across the portfolio of investment products and client accounts compared to 2024. At the advisor-only level, as presented in the supplemental table within the press release, operating income was $5.5 million for the third quarter, up from $1.5 million in the prior year, and the year-to-date period was $16.1 million as compared to $5.4 million in the prior year. These results were driven primarily by the revenue growth we just mentioned. Overall, the company's net income was $0.39 per share for the quarter and $1.05 per share for the year-to-date period.

Please note that our net income or loss is often impacted by swings in unrealized gains or losses associated with certain investments, including digital assets. This quarter, that included an aggregate of $10.1 million of unrealized losses related to our investment securities and equity holdings in our proprietary funds. Those unrealized losses compare to the third quarter of 2024, which had unrealized gains aggregating about $31.6 million. This is an illustration of what we have previously noted for you that may result in seeing volatility from quarter to quarter as a result of unrealized gains or losses in various holdings, including those digital assets, which is primarily Bitcoin.

I should also note that the third quarter of 2024 last year also included the cumulative tax adjustment for converting from a pass-through entity to a C corp, which resulted in an approximately $60 million deferred tax liability and expense upon the adoption of that status. That deferred tax value on the balance sheet changes quarterly based on movements in the underlying cumulative tax differences, including investment holdings or applicable tax rates. From a cash perspective, the company has paid approximately $4 million in income taxes in 2025 year-to-date. From a balance sheet perspective, the company continues to have substantial cash investments, including amounts outside of the consolidated investment products, and we have no third-party debt. Our long-term liabilities are limited to various long-term office space leases.

The company's board recently declared a $0.10 per share dividend, excuse me, $0.10 per share dividend, which reflects a 49% increase from the prior quarter's dividend. This dividend will be paid on December 17th for shareholders of record as of November 25th. That concludes the prepared remarks. I'll turn it back over to you, Murray, just to say a few words before we get into Q&A.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Okay. Thanks, Mark. You have heard that we are required to present our financials on a consolidated basis. I personally think it is important to look at the unconsolidated results, and those can be found in the 10-K, I mean, the 10-Q, I beg your pardon, starting on pages 17, 18, and so forth. Let me just point out a few things there, and then we will talk about some operating matters. You will observe that on September 30th, we had on our balance sheet, this is a consolidated company balance. This is the company equity, not the consolidated equity. We had $37 million in cash. At the same period of time, a year ago, December 31st, we had $14 million in cash. It caused us a little problem, you might recall, because we had these performance fees. Now, it is very nice to have performance fees.

The only problem with performance fees, of course, is that the United States Treasury would like to get its money on an estimated accrued basis, but we do not have access to that money until we have actually finalized the numbers. It was a little bit of an issue, not a big issue, but a little bit of an issue to actually raise the cash needed to pay the taxes that would be applicable for our performance fees, which we are delighted in getting. That impacted last year's dividend policy because we did not want to be caught in that circumstance again. In principle, we had no possibility whatsoever in lacking liquidity because we have a large and liquid securities portfolio. We could have sold securities.

The only problem with selling the securities is they're at gains, and anything we sell at a gain would require even more taxes, which we'd very much like to minimize. I say these things for two reasons. One, of course, is to call your attention to the cash balance and the effect on last year's dividend policy. You'll note that the dividend we just announced is higher than what we paid last year, even though we haven't accrued a performance fee yet, although we might in the future. The second thing is just to make you comfortable with looking at the financial statements the way we look at the financial statements, which is to focus some on the operating company. The operating company for the year-to-date period, and it's the year-to-date period that I would argue is more important than the quarterly period.

The reason it's more important than the quarterly period is because with the possibility of performance fees and with fluctuations in markets, there's only so many conclusions one can draw from the individual quarterly results. That's why I personally like to focus upon the year-to-date results. In this particular case, the operating income line, which you will find on page 18, the operating income line is $16.1 million approximately. That's when my attention is drawn to the financial statement. That's the first number I look at. You'll see below the operating line, there are all sorts of things, entries, and sometimes they're positive and sometimes they're negative, has an accounting impact in as much as we don't really bear the full brunt of negativity because we unaccrue certain tax liabilities.

Just like we share our profits with the United States Treasury, we will also share our mark-to-market losses with the United States Treasury as well. It is never as grievous as the below-the-line numbers appear to suggest, and I think the numbers should be understood in that context accordingly. Now, a lot of interesting things have happened year-to-date. I will call your attention to only several. One is we launched a Japan Owner-Operator ETF in the month of May. That ETF has now achieved $25 million in assets under management. Obviously, $25 million in assets under management is not going to radically change our profitability. Also, some years back, you might recall, we launched another ETF called the Blockchain Development ETF. That ETF has more or less about $20 million in assets under management. Even the aggregation of those, they are not going to radically change our profitability.

However, you will observe that our largest expense, again, from the income statement, is our largest single expense, is sales and distribution marketing. Employee compensation is even a bigger expense, but a lot of employee compensation is marketing-related because the biggest unit we have is marketing. That brings up the biggest problem in asset management companies, which is raising assets. The dilemma that every asset manager wrestles with is how much time will be devoted to raising assets under management because that is the amount of time you're going to subtract from investing portfolios and doing the research. What we'd like to do is we'd like to minimize the amount of time we spend marketing so we can maximize the amount of time we spend doing research and investing so that we can have as good results as we possibly can.

The object of the exercise in the Japan ETF and the Blockchain ETF was to see if we could raise money reputationally. In other words, the products get a reputation. The products were not marketed in a traditional sense. They spread word of mouth. We are looking to see what we did right in attracting the assets. Now, obviously, if they had a lot more assets or alternatively, if new assets come in and they are coming in through that modality, it is an incredibly high-margin business. It is a success, as it were, and we hope to make it a bigger success. If that success does materialize, the conclusion one would draw is our margins would expand, hopefully expand considerably. It turns out that in the first quarter of 2026, just around the corner, we are going to launch another ETF.

We have not really officially publicly announced what it is going to be, but personally, I am pretty excited about it. It has been the same approach, word-of-mouth approach. We will see how far we can come with all that. In addition to margins, the biggest challenge you really have in asset management is the challenge of indexation. Active management in and of itself does not really have the scalability of indexation. Indexation, obviously, since the S&P 500 has trillions upon trillions of dollars of market value, not hundreds of trillions. I think the market capitalization of the S&P 500 is $64 trillion or thereabouts. Obviously, it is a big number. To go from $1 billion of assets to $5 billion of assets is not a problem. That is scalable.

In the world of active management, you're only going to find so many good unrecognized ideas, and you have to realize there's a limit to how much money you can invest in them. Therefore, it's not scalable. Therefore, you have to continually devise new products. That's one of the challenges. We'd like to spend our time finding new investments rather than scaling our existing investments. Towards that end, we've come to a conclusion that we'd like to separate as much as possible raising capital from managing capital. That brings me to another thing I think you'll find intriguing. You'll observe that we brought a company public called Consensus Mining, trades under the symbol CMSG, over the counter. It has a market capitalization of more or less $80 million. Horizon is the manager of the cryptocurrency in their portfolio. What do we mean by management?

Horizon manages the mining of cryptocurrencies, and Consensus Mining, we do something called SCRIPS mining, which is basically to mine two cryptocurrencies simultaneously with one electric current. It's not the place or time to go into detail unless someone asks a question. Okay, so we'll go into detail, but we do that, and as far as I can tell, it's actually pretty successful. The amount of crypto is growing, but it's growing organically. It's not growing from marketing. If the company were to require capital, and maybe one day that time will come, that'll be the prerogative of that company. Horizon is the manager, and Horizon gets a fee. At the moment, that fee is de minimis. Why is it de minimis? Because we're looking to incubate Consensus Mining, but in the fullest of time, contractually, that will pay a substantial fee.

It'll be not substantial in relation to investment management fees as they are known to exist, but will be substantial relative to the de minimis quantities we get right now. We do the exact same thing with Woodland Holdings, which at the moment is 44%+ owned by our colleagues at FRMO Corporation. So there are two businesses. In total, the assets are crypto assets plus the devices that we use to mine crypto assets, and I believe, if I'm not mistaken, the aggregation of the assets is over $50 million. If we were to charge a fulsome fee, that would be a 1%, and the number was $50 million of assets under management. 1% of $50 million is $500,000. At that quantity, $500,000 would come right to the bottom line. There'd be some taxes on it.

You can see we're building the potential for operating leverage, and we're separating to the degree that we can the raising of capital and the managing of capital. Something that just about everybody in the investment management business has reflected upon. To the best of my knowledge, and please correct me if I err in saying this, but I'm not familiar with anyone successfully addressing that question. I think we're up to some really interesting things. We will wait for the new year to see what our new ETF does. Of course, at that time, I will report back to you. We are mining new crypto coins every day, which, of course, adds to our assets under management in the crypto context. I think we're up to some interesting things.

Rather than continue, I'll stop here, and I'll open up for questions if anyone has any questions. As per tradition, we will stay around, answer every question until we've exhausted all questions. If there are some questions, I'd be delighted to answer them.

Mark Herndon (CFO)

Yes, we have a handful of questions already in, and I'll remind our listeners that if you're on the GoToMeeting platform to submit a question, I'll include them in the mix here. Going back to a topic you mentioned a minute ago about the Japanese fund, the question was simply, can you clarify the percentage of the AUM, which, as you mentioned, we publicly disclose the assets under management for the fund. It's on our website. It's about $25 million, which is obviously a small portion of the $10.4 billion of AUM.

I do want our listeners to understand that you can find that information on our website. Is there anything else you want to talk about in terms of the opportunity of the Japanese market overall, just more broadly?

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

I suppose I'd be remiss if I didn't do at least a little bit of marketing since I'm already on the call. Let me put it this way. Virtually everybody in the world of investing is a devotee of diversification. I will share a statistic with you. If you put a European index, and just by coincidence, 13% of the European equity index is pharmaceuticals. If you took the pharmaceutical companies and looked at their revenues and disaggregated them by geographic exposure, you will find that not quite, but almost two-thirds of the revenue comes from the United States.

Now, obviously, the southern parts may seem a little whole, so if 2/3 or almost 2/3 of the revenue comes from the United States, everything else is much less than the United States. Therefore, the United States and the European pharmaceutical companies are the biggest exposure. I think it's legitimate to ask, in a very real sense, yes, of course, they're domiciled in Europe. Are they really European companies? Are you really getting a different exposure? Then that's Europe. Let's turn our attention to Japan. Big companies in Japan: Toyota, Sony, SoftBank, Mitsubishi UFJ. If you look at their financial statements, you will find the lion's share of their exposures geographically come from the United States.

How can it be that one buys an index in Japan with a view to diversifying away from American exposure, and one buys a stock like Sony because it's part of the index, and Sony happens to have well over half of its revenue originating in the United States of America? I think we're entitled to ask, is it a Japanese company or is it an American company? One of the virtues of Japan ETF, I don't want to make it too much of a selling point of the Japan ETF, this is a Horizon Kinetics call, not a Japan ETF marketing call. In any event, what's the point of investing abroad for personal diversification if your revenue is actually going to originate in the United States of America?

This is an ETF where the companies do the lion's share, virtually all of their business, in Japan. For better or worse, this is really a Japan ETF. In relation to Japan ETFs that are largely comprised of the biggest companies in Japan, which are global multinationals, it is distinctly different, and I think it is superior. They have other attributes that I think may even make it further superior, but I'll leave it there, and you can get a sense of what we're trying to do. We're trying to develop very, very unique products, and I think Japan Owner Operators ETF is one such product. I'll leave it there. Okay.

Mark Herndon (CFO)

With respect to our own portfolio, we've had a couple of questions come in, a perennial favorite, to talk about the TPL stock. One of the questions is just simply how much of our AUM does TPL make up. The other is a question about if, as a firm, we have been selling TPL stock. Of course, I understand that with some regulated products from time to time, there may be sales around the mutual fund, or individual clients might provide instructions, but I do not believe anything has changed on that front. I wanted to see if you would expand on our views about TPL in general and our investment.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

The only time we really sell it is where we are compelled to. There may be cases like we have it in a certain portfolio, and the regulations about what your exposure can be, and you have no choice. You have to reduce it. We are not reducing it for any fundamental reasons. If the exposure is too high by regulation, because it's appreciated, we'll have to reduce the exposure, and of course, we're mindful of regulations. We're not selling it for any fundamental reasons. As far as fundamentals go, my opinion, of course, and feel free to disregard, I think the fundamentals are fabulous. I really do. I think they're absolutely fabulous. You can take that for what it's worth, but I couldn't be more excited about the future prospects. You might want to let me leave it there, but if you have more detailed questions, to the degree I can answer them, I'll be delighted to answer them.

Mark Herndon (CFO)

Okay. The next question is also about our portfolio. The commenter has noted the drop in Lamberge stock following a stock offering to their existing shareholders, and they were asking, is that an expression of doubt about data centers or the volume of data centers coming to that area, or just. There are a lot of things I can say about that.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Yes. A lot of things I can say about it. First of all, you may be interested to observe. On October 10th, just for some reason, I just remember that on October 10th, Lamberge traded in round numbers at $49 a share. It is not exactly $49 a share, so forgive me for generalizing, but roughly $49 a share. Some number of days ago, Lamberge traded $85 a share. Today, Lamberge trades at roughly $62 a share. Okay? Clearly, in a little over a month, that is a pretty big movement.

Up and down, that's a pretty big range. Right now, without being too precise about it, we're just in the middle of the range. Interesting enough, today, because the matter was brought to my attention, Lamberge traded as low as $58, as high as $62 and a fraction, even in one day. That's a pretty big range. If it had something to do with the fundamentals, what can possibly happen in a 24-hour time period regarding data centers for any company? If it did happen, it would surely be momentous, and it would be disclosed because it would have to be disclosed. Basically, in my humble opinion, this is just my humble—I underscore the word humble—opinion. I think it's a function in the modern age of how people do research.

Basically, a typical person who does research—I am not like this, but a lot of other people seem to be—they have a screen that they look at constantly, and they have the prices of securities. Some are up, some are down, and some are more or less unchanged. Lamberge is supposed to be active in data centers. They are waiting for an announcement, and they have not seen the announcement yet. I can only imagine the reason. Thusly, perhaps there never will be data centers. If there never will be data centers, then why am I paying X for the shares? They must surely be worth much less, and perhaps I should sell them.

On the other hand, there are days when people come to the exact opposite conclusion, as you can see from the ranges of prices in a very brief period of time, I might add, that I just quoted you. Now, let me impose upon you the following heuristic. Let's make believe, just for the sake of argument—I don't believe this—but just for the sake of argument, there will never, ever, in the course of history, no matter how many millennia, how many eons the world will see until the sun is extinguished and life becomes impossible on the planet, let's say there's never going to be a data center on Lamberge. Never going to happen. Okay? Do you actually think that a company wouldn't prosper anyway? All it would mean is data center is going to be somewhere else.

I'm going to explain where it's going to be in a moment. You can't run a data center without power, and you can't run power plants without water. The only place you can build things of the appropriate magnitude is a place like West Texas. It's happening. Up in Lubbock, Fermi is building its data center and starting with, I think, 1.2 GW. Eventually, in Lubbock, they're going to have over 11 GW. There is, I think, a 2.2 GW plant in Sweetwater, Texas, which is supposed to be energized in April, I believe. I could quote other numbers. A plant or data center is being built for Meta Platforms, AKA, Facebook, in Richland Parish, Louisiana. That's going to start, I think, at 2.5 GW. Either 2.2 GW or 2.5 GW, and it's going to be scaled up to 5 GW.

The reason for mentioning those numbers is as follows. The most data center-dense area of the United States of America is the Washington, D.C., Northern Virginia metro area. If you looked up every data center in that area, every single one, and added them up, it's 3.9 GW. I know because I did it. You just can't build a 2 GW-3 GW data center in the area that you can't get the electric power. You could get electric power. You couldn't get the water. I'll explain why you need the water in a second. You don't need the water so much to cool the data center. You need the water to generate power. You need a site.

If you're going to build these magnificent structures and the power plants to go with them, you have to have a site with a lot of water and not a lot of people. Over the course of civilization, as you probably know, people, wherever they are in the world, had a tendency to move where there's a lot of water. There aren't too many places in the world where you can get water and you don't have to deal with people. One of the few areas is West Texas. Now, before I go more into water and why you need the water, I'm going to give you one other physical fact so you understand this question. Let's say, merely for the sake of argument, I decided to build a 2-GW data center. Okay?

I'm going to give you a hypothetical question, and I'm going to answer it, so it's rhetorical. If I'm going to build a 2-GW data center, how many GW of power plant do I need to service it? Okay? If I asked virtually any person, they'll say, "Well, if you have a 2-GW data center, you obviously require a 2-GW power plant." As reasonable as that answer appears to most people, that is wrong. Why is it wrong? For two primary reasons. First thing is that no power plant can possibly be up and running 24/7. It has to go down, sometimes either for scheduled maintenance or unscheduled maintenance. When it goes down, you have to have backup power. Since you have to power a data center, it can never be down. At minimum, you need another 2 GW.

Therefore, a 2 GW data center will require 4 GW of power. Okay? Now, okay, I build a 4-GW power plant, and it might go down on a moment's notice. I have to have both things running simultaneously. One is the power plant that's feeding the data center. The other is the so-called spinning reserve. Okay? There is other backup power needed. You have to deal with another issue, which is called PUE, Power Usage Effectiveness. What does that mean? That means if I had a 2 GW data center and I had one power plant, 2 GW, I could not possibly energize that power plant. Let's forget about the issue of the redundancy for backup. Why?

Because the amount of power you're drawing, the wattage you're drawing, is so enormous, even though the power is only moving a short distance, that much of the power is being lost as heat. Basically, the electrons in the wire are hitting each other, and they're hitting the walls of the cable. A lot of power is being dissipated because of that, and that's expressed as heat. It's the same concept as when you charge your cell phone, and you touch your phone, you touch the charging device, you'll see it's warm. You're not using all the power that comes from the outlet. It's being lost. It's the same concept, except it's not a small device. It's 2 GW. The general number you'd use is the coefficient is 1.58. That's your power usage effectiveness. What does that mean?

That if you had a 2 GW data center, you have to draw 1.58 times 2 or 3.16 GW. You need 3.16 GW to energize a 2 GW data center. That's how much power you need. It's just enormous. Okay. That answers the power question. Now, the water. Why in the world do you need water? Because people think, "Well, you need water to cool a data center," but they can get around that. They have various ways of mitigating the amount of water they need, and that's all true. I won't go into details because we're not in real agreement, but we're in general agreement. You don't need a lot of water to cool a data center, even though you do need some. The power plant.

The power plant, whether your power source is coal, which you wouldn't use, or it's nuclear, or it's natural gas, all those three things are thermal. What does that mean, thermal? It means whatever your fuel source is, you're generating heat to boil water. You generate heat to boil water. What happens when you boil water? It turns into steam. What happens to the steam? It goes through the turbine and spins the turbine. Without explaining how the rotors and the stator actually generate power, let's just leave it at power is generated. In order to generate it at a constant rate, the steam has to move through the turbine at a constant rate. How do you ensure that it's at a constant rate? You have to basically condense the steam on the other side of the turbine.

If the temperature is lower on the other side of the turbine, there is going to be a temperature differential known in physics as temperature gradient. The steam is going to move from the hot area to the cold area at a constant rate, the same way water would flow through a pipe at a constant rate. That is how it is done. How in the world are you going to condense the steam? There are only two ways to do it. Way number one, what some people refer to as an open-loop system. How does an open-loop system work? It is open, meaning you expose the water to air. You expose the steam, rather, to air. It is the ambient temperature of the environment. Because steam is at 212 degrees Fahrenheit, when you expose it to air, its temperature is going to cool. It is going to condense the water.

hundred percent condensed water, some of it's going to evaporate. That's why when you look at a power plant and you see this big smoke stack and you see white smoke coming out of it, that's steam. You're losing some of the water. You need this water to be replaced. At what rate do you have to replace the water? The number is so big that I won't even tell you what the number is. I'll give you a homework assignment. Just go to the website of the American Electric Power Association and look it up. It's a big number. Even though you're condensing 90% of the steam, maybe more than 90% of the steam, it's still a big number. You got to have a lot of water. That's the way the thing works. What's the other way of doing it?

The other way of doing it is a closed-loop system. So basically, the steam never leaves the pipe. The steam is routed back to the front of the turbine. Okay? It has lost some of its heat energy, so it has to be fired up again through the boiler. One, you lose some of its energy just because heat is dissipating through the pipe. You have to keep boiling the water. You still have to condense the steam back into water. Now, it is not going to evaporate because it is in the pipe. How are you going to get the steam in the pipe to be recondensed to water when it is not exposed to air? What do you do? You pour water on the pipe because the water is what is known in physics as a heat transfer fluid.

The water on top of the pipe turns into steam and it evaporates. You need a lot of water. The operative thing is, no matter how you do it, everyone's looking for an announcement. I put this 2 GW data center on my property. The operative thing is the water. Believe me, there'll be announcements eventually. The operative thing is to allow for the water. You can't draw the water out at a rate in excess of the natural refresh rate, or you'll destroy the aquifer. If your neighbor happened to be fortunate enough to have a data center on your neighbor's property, your neighbor is unlikely. By the way, it's not the data center. It's the power plant that counts, not the data center, because it's the power plant that's using the water.

Whether it is on your property or not, if it is on your neighbor's property, you are sure your neighbor is getting the ground rent. The real money is the water. Hence, your neighbor cannot draw the water at greater natural refresh rate because your neighbor does not want to destroy the aquifer. Your neighbor is going to have to collaborate with you and take water from you anyway. You are all going to be sharing it. I just told you this. If people realize that, the stock will probably not be fluctuating the way it does. Unfortunately, we do not live in that universe. The stock price is what the stock price is. You see what their expectations are, and you see the reality. In my humble opinion, I underscore the word humble, of course.

They ought not to be as excitable as they are at the moment is, I suppose, my concluding remark on that subject. You can follow up and ask me more questions on that if I have not been sufficiently concise and clear. Okay.

Mark Herndon (CFO)

We are going to stay on the forward-looking investment horizon here a little bit. The question is then, over the next three to five years, do you foresee the global AI buildout leading to net inflation due to CapEx spending or net deflation due to productivity gains? Do you have a view on that?

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Neither. Neither. I just wrote about this. I do not think it is out yet, but I will give you my investment thesis. The biggest exposure in the S&P 500, it is obviously technology stocks. Whether it is Microsoft or Meta Platforms or Google a.k.a. Alphabet, or whether it's NVIDIA or maybe Broadcom or maybe AMD, you can add it up. You can see what it amounts to. It's the biggest exposure. If you simply take the earnings of the big data center companies, they just reported them a week and a half ago, get the cash flow statements. What you will see is you will see very large capital expenditures. They're connected with data centers, all right, but they're not building data centers. Just like if you're a big company, you want to rent office space. You don't build the building, but you're still responsible for buying your furniture. You're still responsible for buying your computers and communications devices and electronics and office furniture and so on and so forth. Same with data center. The big companies, basically what they're buying is they're buying electronics. They're buying the service.

From whom are they buying the service? NVIDIA, AMD, Broadcom, etc., etc. Okay? If the data centers are going to continue to be populated, the capital expenditures are going to be high, and your free cash flow is not going to be very high. Because your free cash flow, your earnings might be higher than what they otherwise would have been, but your free cash flow is going to be lower. That is going to affect the valuation in a negative way, even if the earnings go up. You might say, "The capital expenditures will go on only for a brief period of time, and then they'll cease because the data centers will be built." There we have a problem. That problem is only—I've written about this. This problem is only dawning on the market right now.

That's why the market is so tenuous. Two possibilities: either that assessment is correct or it's incorrect. Let's say, just for the sake of argument, the assessment is correct. Amazon, Microsoft, et al. will significantly reduce their capital expenditures, but they're buying the equipment that goes into the data centers. They're going to reduce their capital expenditures. It must logically follow that they're going to reduce their purchases from NVIDIA and Broadcom and AMD, et al. That segment of the market will fall in value. We'll probably offset because the capitalizations and the ratings of the S&P are just about as big as the companies who are buying that stuff. If indeed the capital expenditures are going to decline, it's making a problem for a subset of technology, which is big enough because the valuation is very high, to create a real market problem.

Possibility number two, the capital expenditures keep going up. Why would they keep going up endlessly? A simple reason. Because NVIDIA asserts, and they assert with very good reason, that they're doing what Intel had done a quarter century ago, which is they're obsoleting their equipment every 24 months. The current iteration, the Grace Blackwell 200 GPU, graphics processing unit, it will soon give way to the Rubin Vera equipment. You have to get rid of your Grace Blackwell 200s and buy new stuff, just like Intel was doing a long time ago. If they intend to do this, which they obviously do intend because they do say that, you always got to stay with the state of the art.

If you always have to stay with the state of the art, the state of the art is going to be obsoleted every 24 months, and your capital expenditures will never end. You'll never really have free cash flow, and that's horrible. You might say, "Well, in that case, if that proposition is true, I'll just go buy NVIDIA." NVIDIA has a real intriguing advantage relative to what Intel had. Intel, as you might recall, had the problem of, yes, they obsoleted their microprocessor. Yes, the new generations of PCs, personal computers, will be required to include or embed the new Intel microprocessor. They had to build a whole new wafer fabrication plant. What good was it? They had more earnings, but they had less free cash flow. To what end?

NVIDIA, you will undoubtedly argue, does not do that. Because NVIDIA does not make chips. Yes, they make a few chips on an experimental basis, but they do not mass-produce chips. They have some minimal wafer fabrication facilities, but the chips are made by Taiwan Semiconductor. Only their wafer fabrication facilities will be obsolete. Therefore, NVIDIA is in an unavailable position, but not really, because you think Taiwan Semiconductor will live without a payback on its capital. For the privilege of making the chips, they will charge accordingly. That will create a problem for NVIDIA. You see, no matter which way you go, there is a logical contradiction in the entire movement. Logical contradiction dates back, and this will explain our horizon exposures. The logical contradiction goes back 10, 12, 15 years.

Because the premise was you can build a gigantic business, like the technology business, without using resources from other segments of portfolio. And for a while, they were actually doing it. It is a very ahistorical view. For example, historically, the automobile was invented. You could have been a big believer in the automobile more than 125 years ago, and you never had to buy an automobile stock. You could have bought Exxon or some other oil company, and they needed the gasoline. You might have concluded, "I'm going to the best automobiles by buying oil companies." Exxon actually would have been a good choice. Then it was called Standard Oil of New Jersey, but nevertheless, you get the idea. There are other things you could have done. You did not necessarily need to buy the automobile companies. Other segments of the economy prospered.

For example, they had Bill Roads. They were gravel companies, cement companies, construction companies. Some of those were actually very prosperous. You could have had a whole portfolio investing in the evolution of the automobile and never bought an actual automobile company. Strangely, that did not happen in technology. It did not happen in technology because the iPhone invention, you could run the iPhone, the smartphone, all these brilliant, beautiful devices on the existing fiber optic cable of telephone systems. For 10 or 12 years, you could expand the business with minimal capital expenditure because you were building on an existing network. Netflix, for example, Netflix, they're nice and Netflix uses 25% of the internet bandwidth in the United States of America. They do not have a lot of capital expenditures because the fiber optic cable is already in the ground. That is a very ahistorical, unusual situation.

Now we're coming into the different world, the different world of high-performance computer. Note how I do not say artificial intelligence, but high-speed, high-performance computing, because that's really what it is. We're going to need other things. We're going to need water. We're going to need natural gas. There's only a handful of ways to get them. At Horizon, we got them. We're staying with them. It is just a matter of time, the world is moving in the direction to make these, all of them, extraordinarily lucrative investments. You do not even have to do anything. All you need to do is think it through. Perhaps, just perhaps, I may advance the proposition. Maybe that's why Lamberge, for whatever reason, was actually up today, not down.

Anyway, I'll leave it there, but I'm willing to elaborate if someone wants to propose a more involved question. Okay.

Mark Herndon (CFO)

I may have glossed over a question earlier. This is just a factual question of what percentage does TPL make up of our AUM? I can just let our investors know. We disclosed that last year in our 10-K, that it was roughly 41% as of December 31st, 2024. That's been one of our risk factors. It's come down a bit since then, but I don't believe we've put out an updated number. Generally speaking, that's about where it is. Right.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Basically, yes, because when we get new assets, yes, because let me just elaborate, if I may, when you get new assets from management, we obviously can't buy, and we wouldn't buy TPL at that allocation. New assets diluted. For example, Japan ETF, say what you will about it, positive or negative, does not have any TPL in it. Similarly speaking, the blockchain development fund has no TPL in it. As we raise new assets, there are different investment products. They do not have any TPL. It gets diluted over time, even without selling.

Mark Herndon (CFO)

We have a series of questions now that are turning back to just the Horizon Kinetics holding company stock. This one is very simple. What is your single highest priority for deploying free cash flow right now?

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Single highest priority. Obviously, we have a 70% payout ratio. We are taking 70% of our net income, and we are paying it out to shareholders. I think it is fair to say our biggest priority is rewarding the shareholders. Yep. Along that same line, there is a question about government structures.

Mark Herndon (CFO)

What government structures are in place to ensure minority shareholders can fully participate in the long-term compounding of HKHC's capital allocation strategy, which I suspect you're going to talk a little bit more about dividend there.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Dividend is one thing. Obviously, everyone gets to participate. We're long-term investors. We encourage people to be a long-term investor. We're open, but I think we're open. I'd like to believe we're open. I don't have the answer to your question to hand. Usually, I do, but if I don't, I'd be delighted to get the information, engage with you, and get you an answer. We're pretty much an open book. We have no plans to take the company private. If maybe that's what you're thinking about, that we might take the company private again, we're not planning on doing anything like that.

There are no plans to do that. You might observe in the last open period, I personally bought some stock. I can't buy stock today, obviously, because I'm having this call. We'll see what I do when the next open period is. I don't know what else to say. It's available for anyone. It really welcomes long-term investors.

Mark Herndon (CFO)

Yeah. It's really the same structures as any other public company, right? Okay. The next question, a little bit more specific back to our financial statements. Again, this one is a little specific to our lease obligations. The question is around new or extended lease obligations that are listed in Note 10. Is that all from New York office space if I may, just to put a couple of clarifications around that for our listeners, our lease obligations are for office space broadly over multiple locations, not necessarily just in New York City. What you see in Note 10, the contractual obligations table within Note 10, includes the run-out of existing New York City offices through early 2027, as well as two other previously existing office space leases and one new 10-year lease in Connecticut for additional office space. While we've disclosed additional office space agreements, which are in both New Jersey and New York, those leases have not yet commenced. We haven't taken occupancy for those locations yet. Those are omitted from what you see in that table in Note 10. Rather, the broad terms are disclosed in the narrative sentences below that note.

I will let you know also that that lease obligation in New York City is for different space than what we are currently in. The new lease is for 15 years, which is why you predominantly see this gross number that you may see in the note total payments over that long-term time frame. I'll emphasize to you now that the cash payments that we will be making eventually are actually less than what we're currently paying in New York City. That is a very long setup, but I'll just, Murray, I'll turn to you to ask. Would you like to comment on how we plan to use that New York City office space or in general?

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Yes. Basically, we came to the conclusion that if you've been to our offices, we're in three separate floors. That creates one problem, and then there's a second problem. The first problem is it's very inefficient. We have three lobbies. We have three reception areas. You actually pay for the elevator entrance. There is a lot of wasted space. It is very inefficient. It would be a lot more efficient if we could be on one floor. Part of the reason was to save money by going to one floor. The other thing is the current lease, we are responsible for paying for all sorts of ancillary services that in another location, we would not be responsible for paying. That is the savings as well. Basically, in terms of the usable floor space, we would have a better arrangement. In terms of the ambiance, it is better. In terms of the monthly obligations, you pay monthly, it worked out to be 35% cheaper. Everybody can have a better experience. We save 35%.

It seems like everybody should be happy. Speaking as a very large shareholder, it made me happy. That is where we are going. We also have another issue that the new space is going to solve. You might be aware, I myself do a lot of meetings. We call round tables and other things where I invite investors to come in. The existing accommodations do not normally allow for the attendance that we would otherwise have. We have to limit it. A lot of people like to come in and hear what we have to say, and it is kind of foolhardy to not let them do it because it is also a source of new business among other things. We needed to find a space where we would not be so limited. We found such a space, and that is what is happening as far as that goes. Have I left anything out more?

Mark Herndon (CFO)

Yeah. We have a couple more. We've had a couple of questions come in around the mechanics of our disposal of the consumer products division. This person references specifically the Scott's Liquid Gold sale. I did want to just spend a minute to set this up as well. Many people get confused about the Scott's Liquid Gold corporate entity that we merged with in 2024, which is actually different than the namesake brand that they had disposed of well prior to our transaction with them. The two brands that we had were called Kids n Pets and Messy Pet. Due to their relatively small size, the revenues associated with those brands were reported within our other revenue line in the past. As a group, again, we called it the consumer products group as opposed to the individual brands.

The question was about, were there multiple bidders, and can we share more details about the sale, and where the $2.5 million fair value was recorded on the balance sheet? If you'd like to comment on it broadly, I can also give a couple of specifics if you'd like.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Yeah. Basically, we did get some cash, and then we're also getting a long-term royalty. We like royalties. Long-term royalty gives us some upside. Obviously, we're never going to have the scale to be in consumer products. Basically, disposing of inventory, getting some cash, and disposing of the ongoing expenses or limiting the ongoing expenses, which we did, and getting the long-term royalty. Now it's just cash flow over a very long period of time. You can maybe elaborate on that, Mark.

Mark Herndon (CFO)

Sure. Be happy to.The one thing, just mechanically, the $2.5 million fair value that you referenced is in the other assets sort of down the balance sheet line a bit. I would say one of the things we talked about is just that there's some parameters around the royalty arrangement that this percentage, we have a little higher percentage in the early years, a little lower in the outer years, and a minimum of $1.5 million and a maximum of $5.25 million over this long-term period. I think we came to the conclusion that the additional focus that this buyer can bring to it, that that is going to bring more to us from those brands in terms of royalty income than we would have achieved by just continuing to own and operate it ourselves as a standalone operation. I think that's the last question specific to HKHC.

There has been just one more that came in while we were talking about that. Yeah. Let's do it. Around sorry, just looking for the question again. AI as currently defined I'm sorry. It says, "Murray does not believe we have AI as currently defined by the industry. Do you foresee any superintelligence being created in the near future? Are we on that trajectory?"

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Okay. I don't have any more. Let me just explain. Obviously, it's more than just Horizon Kinetics. In the colloquial, artificial intelligence is taking the mean, the average person, machines that think like human beings. The problem is it's very difficult, and many would argue it's impossible to get a machine to think like a human being. Why is that true? Because a human being, even though human beings do many, many foolish things, a human being has the ability, at least in principle, to recognize the limitations of human thought, which is not to say to recognize the limitations of that individual's human thought, but the limitations of thought itself.

For example, without giving you a treatise on epistemology, there are certain inherent contradictions in thought. For example, there is what is called the Cretan-Liar Paradox, also known as the Epimenides Paradox. It goes something like this. Epimenides the Cretan says all Cretans are liars. He is a Cretan. Therefore, if that statement is true, he must be lying. Therefore, all Cretans are really telling the truth, but everything he says is a lie. It is an inherent contradiction. How do you resolve it? The answer is it is unresolvable. In mathematics, you find the same thing.

You find something called Gödel's incompleteness theorem that was discovered that you can't complete. There are certain mathematical propositions that you cannot completely prove. So proof that there are certain things that can be proven. You could have in mathematics also the Cantor infinity paradox. No matter how big a number you come up with, it's always going to be bigger because you can add one to it. If you can add one to it, why can't you add infinity to infinity? Why can't you make infinity squared or infinity cubed or infinity to the hundredth power and so on and so forth? You never get to the end of it. How do you ultimately talk about a finite universe if numbers are infinite? It goes on and on like that. When you get into really complicated questions, you have to deal with that.

You also have to deal with it. I gave an example of this at a recent round table, so I'll recapitulate it for your amusement. Let's take a more these are all abstruse, very refined questions, and maybe we shouldn't even be talking about them. Let's do a simple exercise in what some people call artificial intelligence. I will describe the exercise, and I'll tell you where we got it from. It's a ChatGPT exercise. The exercise is as follows. You ask a computer to make a list of all the books written by Winston Churchill. Okay? We're not talking about the Cantor infinity paradox or Gödel's incompleteness theorem or the Cretan-Liar Paradox. There's a certain number of books that Winston Churchill wrote, and you got to look it up.

The problem came up because at the President Franklin Roosevelt home in Hyde Park that I visited not long ago, they redid it. In the study, because President Roosevelt was friends with Winston Churchill, they decided to put in the study all the books of Winston Churchill. That's a cool exhibit. The trouble is that there were books in there that said they were written by Winston Churchill. They were not written by Winston Churchill. If you keyed in ChatGPT, give me a list of the works of Winston Churchill. Until recently, they may have corrected it because it was pointed out to them. Until recently, they gave you works that were not written by Winston Churchill that they said were written by Winston Churchill. How can I make such a mistake? We're not talking about complicated things.

Because Winston Churchill was alive, and Winston Churchill was writing books. It just so happens, there were two Winston S. Churchills writing books. One was Winston Churchill, the orator, the statesman that you know. The other was an American novelist known as Winston S. Churchill, who lived in the city of Boston. In his era, he was a best-selling novelist. Now, how is ChatGPT to know there are two Winston Churchills, not one? Interestingly enough, Winston Churchill, during his life, corresponded with a novelist. He wrote, "I'm going to paraphrase a letter that he wrote." The letter was written before Winston Churchill became prime minister. He said, "I intend one day to be the prime minister of the United Kingdom. It wouldn't be great if you ran for president and you won, and we could meet as Winston S. Churchill each.

You as the President of the United States of America and I as the President of the United Kingdom. That would be pretty intriguing." Alas, that is not the way history was written. ChatGPT does not know that. You see the problem is every nuance of every bit of data has to be pointed out to the computer. All that data has to be cataloged properly, and all that data has to be manipulated. Somebody is producing a query. That is why the typical ChatGPT query takes 10 times the amount of electric power than a simple Google query. You will say to me, as a rejoinder, as most people inevitably do, "One day they will figure out a way to use less power for more elaborate queries." You will not be right if you say that. Why?

Because if you look inside a computer, a computer is basically just a big electron magnet. The data, it's expressed in binary forms. It switches on and off, zeros and ones. In other words, the data is simply electrons. So you can't process more data by using less electrons because data are electrons. You can't use more electrons. The question is, for society, if you want to have those privileges, which I personally don't need them, but I'm a minority of one, everybody else seems to want them. If you want them, if you want autonomous driving and you want to watch movies at 3:00 A.M. and you want to write book reports on Shakespeare without having read Hamlet, that's fine, but you better be prepared to use a lot of electric power. This world is running out of it.

As I said, I'm a minority of one. I don't need the stuff personally, but everybody else seems to want it. Since my vote doesn't count for anything, it is what it is. It's going to be the way it's going to be. You know what the outcome is. Sometime, not too distant future, the best guess is sometime in the year 2028, we're running out of electric power. There's nothing anybody can do to stop it. That's where we are. If it went the other way, just for the sake of argument, mind you, it went the other way, it went the Murray Stahl way. If it went the Murray Stahl way, instead of reading a book on a device, you might just buy the book and just put it on a bookshelf. That doesn't use any electric power.

Problem with that, to show you that everything has its problems, believe it or not, there are 5 million books published in the world every year. How would you ever build a library to contain them all? Even if you could, how would anybody be able to use it? It's just too many books. You can't flip through them. You can flip through some, but basically, you have to have some sort of electronic means of surveying them and extracting data from them that's useful. Therefore, there's no alternative but to go to high-performance computing, which everyone calls artificial intelligence. That's not artificial intelligence. It's merely data sorting. I hope—sorry for the elaborate answer, but it required it, and I hope you found it helpful with respect to your question.

Mark Herndon (CFO)

Okay. That's great. I do not have any further questions coming from the web or sent in advance.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Okay. Excellent. I take it that that is, they say in the movie business, a wrap. Is it a wrap? I think it's a wrap.

Mark Herndon (CFO)

I think it's a wrap.

Murray Stahl (Chairman, CEO, and Chief Investment Officer)

Okay. Good. In that case, I'm going to thank everybody for listening. I thank you all for your support. Of course, you know what happens that we hang up this phone and someone's going to think about something they wanted to ask but have got to ask. Please do not hesitate to contact us if such an event occurs. If you have a question, we will get you an answer. Until then, I know, of course, we'll reprise this in about 90 days. Until that time, thank you so much for attending. We look forward to seeing you at the next Horizon Kinetics shareholder update. Thanks so much. Bye-bye.

Thank you.