Quantitative tightening is officially over. You'll start seeing some quantitative easing. The repo market's going to need help. There are a lot of areas where I think you're going to see money being injected into the economy. The biggest constraint in the AI business is power. And the fastest way to get power for an AI site is to convert a Bitcoin mining site. What you're going to start seeing is AI data centers learning how to be curtailable loads. And I think that if anything will create more competition with Bitcoin miners. November was one of the weakest months for Bitcoin since 2021. We're going to find out why and what's next for Bitcoin miners. Are miners dumping Bitcoin to transition into AI? Has that been one of the causes of the sellout? We'll find out why and what's next. Fred Thiel joins us today. He's the CEO of Marathon Digital Holdings. Welcome back, Fred. Good to see you.
A lot of interesting activity to talk about, starting with market topping and market selling of Bitcoin over the past couple months. As you're aware, November has been a weak month, October as well, but in particular November. Let's just start by asking you what happened here and then we can dive into the details.
Well, those of you who may have listened to our Q2 earnings call may remember I made the comment that I felt the market was getting a little frothy. We had seen huge tailwinds around institutional buy-in, inflows to ETFs. You had a lot of expectations of Fed rate cuts, etc. And the Japan carry trade was still carrying along quite nicely. And then along came November and you had a couple of things hit back-to-back. One was a Fed rate cut became all of a sudden low likelihood, Polymarket started scoring it way down. That caused a risk-off. There was the scare around AI assets and all this debt, all these things going on around the AI circular economy and people getting a little tenuous on that. And that caused even more risk-off. And then you had the rate hikes in Japan that made the Japan carry trade, people needing to unwind that. In addition, we've had all along since late summer, whales, older whales who have wallets that are arguably 10 to 13 years old, even older in some cases. And by the way, these are wallets that are highly susceptible to quantum attack because the public keys are well known and recorded, and they don't have any of the Taproot protections in them. And so you started seeing whales either selling off Bitcoin that they had or actually just transferring it into ETFs, which looks like a sale. When you see an old wallet that hasn't moved in many, many years all of a sudden move, people freak out. And so there was this compounding effect of all of that. And when you get a risk-off environment, Bitcoin operates a lot like equity today, the way people manage it in their portfolios. And you started seeing this kind of, between Jim Chanos shorting strategy, all these things kind of layered on one over another, and then you started seeing highly leveraged positions unwind. We'd had a couple of times in October a large whale who took a huge short position on the perpetual markets and made out really well with that October dip. And I think you just got other people kind of piling on in doing that trade. And here we are, we got a correction. People are still kind of wary as to what's going on. Obviously the whole digital asset treasury company model seems to be being called into question. Strategy traded below its NAV, you've seen a number of these other ones, Nakamoto, MetaPlanet, etc., drop precipitously. And then you had American Bitcoin when the lockups expired, the share price at one point was down I think 50%. But so you had all of these events that just created a lot of fear. You had withdrawals from the ETFs. And so here we are, a few weeks later, and I think we found some stable ground, stable footing. I think the Fed rate cut is pretty much priced in at this point. The question is what happens next year. If you look at the macro environment, obviously money printer go brrr is kind of the least of what you're going to see. I think quantitative tightening is officially over. You'll start seeing some quantitative easing. The repo market's going to need help. There are a lot of areas where I think you're going to see money being injected into the economy, especially as we're going into a midterm election year that the Republicans absolutely want to win. I'm pretty certain we're going to see a lot of stimulus going into the market. Remember, for the deficit and for the debt, having a dollar that's declining is a good thing. Having growth growing dramatically is a good thing. The problem with all that is inflation. And how do you keep that under control? You're now starting to see jobless rates tick up, but we're still at a little over 4% unemployment only. So I think there's a period of time during which you had a lot of uncertainty. The government shutdown didn't help it at all. Lack of data didn't help it. And now I think we're on a little bit more stable ground. And we'll have to see what happens here between now and the end of the year. I think all of the tax selling and all that has already happened. So we're pretty much in this coming up to the Christmas lull where we're not going to see anything and then it'll be game on again in January.
I interviewed Matthew Seagull from Van and this is what he had to say a couple weeks ago. I'll play you just 30 seconds. Take a listen. We'll comment together.
It's a pretty expensive upgrade cycle, but the returns over the cycle are better. So they're basically selling Bitcoin to fund this pivot. And when credit conditions tighten, then the cost of the debt that they need to repurpose this facility goes up. And if Bitcoin sells off because credit conditions are tightening, then they literally need to raise more debt at the same time. So these are really levered bets on AI. And when AI and Bitcoin are correlated, it's going to be doubly bad for the stock. So that's what we've seen here in the last month.
He was talking about miners. Can you comment on that?
Yeah, I think if you have a lot of Bitcoin and you're borrowing against that to invest in AI, and as an investor, not as a miner, and you see those profiles go negative, then yeah, it's going to cause a lot of issues. I think as you look at miners investing in AI and borrowing against Bitcoin, that's just a challenging situation. I think if Bitcoin price is depressed, Bitcoin miners really need to be focused on mining Bitcoin and then looking at how they can leverage the assets they have, the portfolio of assets they have, and converting those into higher value assets where you could do AI. And there are multiple ways to do that. You can either go to an HPC developer and say, 'Hey, I've got a site. I've got power. Take this, turn it into something, pay me a long-term lease.' That's what a number of the miners have done. Some miners are saying, 'Hey, I'm going to go build a building. I'm going to put some raw metal in it, meaning I'm going to go buy GPUs, and I'm going to run those and I'm going to rent that out to somebody.' And then there are people who are just doing land and power deals with developers from a greenfield perspective. I personally think that the miners that have appropriate sites should look to converting them into high value AI sites possibly. There's also a hybrid approach where you can mine Bitcoin and operate AI at the same site if you have sites that are big enough and have the right type of internet connectivity and access to other resources. And then there's also the model that we talked about on our Q3 earnings call, which is you start running inference AI yourself as a miner. We do it in the same containerized modular format that we do our Bitcoin mining. We don't have to build big buildings. We don't need water cooling. It's all air cooled. And so we're beginning to put our toes in the water and move towards a vision that we have regarding inference AI. We think training is something that should be best left to the hyperscalers. But as you look at inference AI, there are other things that become important. You need to be able to offer private cloud because if you are running inference AI, you're running a model and you're provisioning it with data. That data is like gold to you, so you want to protect it. Private cloud becomes very important, and one of the reasons we made our investment in this company in France, Exion, that's going through final regulatory approval in France, is essentially here's a company that was developed within EDF, one of the largest energy producers in the world, to operate and manage their data centers that run their nuclear power plants. They have amazing technologies around managing and operating private cloud, data sovereignty, data protection. And it's something that we believe is a core component you need to have if you're going to offer inference services. And it allows us to provide a more rounded out solution to the marketplace. And we look forward to talking a lot more about that once it passes regulatory approval here and as we lay out our full roadmap of what we intend to do in the AI space.
The global financial system is rapidly moving on chain and today's sponsor Cool Wallet is a name worth watching. Cool Wallet builds next generation cold wallets designed for a world where real world assets, payments, and savings live on the blockchain. Their flagship credit card-sized hardware wallets combine CCAL6 plus secure element protection, mobile first usability, and a unique 2 plus 1 factor authentication system that requires your phone, your biometric ID, and your card to approve every single transaction. So here's a Cool Wallet Go card. It's an easy to use on-the-go card-sized hard drive that stores your private keys and literally fits inside your wallet or purse. Activating the card is simple and takes about 30 seconds. All you have to do is download the Cool Wallet app and follow the onscreen prompts. Let me show you now. Activating a new wallet is very simple. Simply go on the app, click on add wallet, and it's going to ask me if I have a card. Yes, I have a card. Every time you order a Cool Wallet Go product and package, it comes with two cards, primary and backup. It's going to ask me which card type. In this case, Cool Wallet Go. Notice how it pairs with NFC. Click on that. Create a new wallet. Activate primary card. Either of these cards can be your primary card. Simply scan the card by tapping it to the back of your phone. Success. And it's important to know that you need the primary and backup cards to be able to fully activate the entire wallet. So both of them have to be active. Now create a new pairing password. This is important for when you need to recover your wallet in case your cards are ever lost. Generate by card. It's going to ask me to generate a private key. Tap my primary card. Start setup. Now it's pairing my private key to the primary card. Now that both cards are paired with the private keys and activated, the wallet is ready to go. As you can see, it's a brand new wallet, so it has no funds in there. Every time you put money in there and you want to transfer funds to another wallet like this, for example, it will require you to scan and tap the card to the back of your phone as a security precaution. That's just one of the benefits of using Cool Wallet. It's very secure. Remember how I said to remember your pairing password? Well, if ever you lose the card, go to settings, go down, replace primary card, and it's going to ask you to enter your pairing password. That will start the process. So once you've activated the wallet, you can use the app to swap to other tokens and cryptos. All you have to do is first click on the manage icon on the app, add the crypto that you want, then go to the homepage and click on swap. And the crypto you just added should be there. You can also send cryptos to another wallet, making peer-to-peer transactions instantaneous and frictionless. There's also an earn function that allows you to stake your holdings to generate yield. Cool Wallet also recently launched a breakthrough fiat on-ramp feature. So instead of going to a centralized exchange first, you can on-ramp directly to Cool Wallet. Users can create a bank account under their name and a fiat currency of their choosing that's available, wire the funds in, and have those dollars automatically convert into a stablecoin that's onchain directly into your wallet within minutes. With support for 40 plus mainnets, staking, lending, NFTs, and advanced clear signing protections against phishing and malicious contracts, Cool Wallet is positioning self-custody for mass adoption. To learn more, go to the link down below or scan the QR code here. And if you do use my code, you'll get a 5% off on your first purchase. I'd also encourage you to check out my interview with Cool Wallet CEO Michael O for more details on the product and why we need self-custody today, and he also explains in great detail how the security functions work. Link also down below. Okay. So Mara is one of the world's top miners and also according to this website bitcointreasury.net, it changes regularly, but currently as we speak today, apparently Mara Holdings is number two in the world for top 100 public Bitcoin treasury companies. So this is the chart that they've shown on their website. Let's comment on first of all whether or not the data is correct in the sense that the BTC holdings from Mara Holdings has increased steadily over time over the last year at least. And it doesn't indicate any selling from your balance sheet. So you've increased your holdings from roughly 44,000 to now 453,000 BTC. Is that correct?
It's actually a little higher, but yeah, that's more or less.
But you haven't sold throughout this year. The general direction of the holdings has gone in the direction that the chart indicates.
Yeah, I mean, we announced in our Q3 call that we would start selling some of our production Bitcoin, which we have done, but we net increase our balances every month anyway.
The transition to AI, I'm going to bring this up one more time. According to this article here, 70% of top miners are pivoting to a $20 billion AI market. Bitcoin mining profitability plunged to record lows in late 2025 as the hash rate dropped below $35 per petahash per second. I believe your hash rate has gone up. Correct me if I'm wrong. But I'll let you talk about that and this trend overall. 70%. I wonder why Mara isn't perhaps following this trend closely in the sense that you're selling Bitcoin to fund a transition to AI. Maybe you don't need to or maybe you've taken a different direction. What's your approach?
Well, I think a couple of things. One is we're certainly not cutting back our hash rate. And if you look at global hash rate, global hash rate certainly hasn't slowed down other than in the past 2 or 3 weeks with the price of Bitcoin dropping, you've seen some older mining equipment get shut down. But that doesn't mean that miners have necessarily shut down or converted sites. I think most miners that are chasing the AI HPC bug, like Galaxy, Riot, etc., are looking to take capacity that they haven't brought online for Bitcoin mining and converting that into HPC AI capacity. The problem is if you have an existing mining site, you have to deconstruct the mining piece and then construct the HPC piece. Unless you do things like we do, which is modular using containers. We don't build buildings. We don't do liquid cooling systems with direct on-chip cooling for AI. So it's a different model. If you look at what the vast majority of our peers are trying to do, it's either developing a new site, greenfield site, and they say, 'Hey, I've got power. I've got this location with land, Mr. Hyperscaler, come, let's cut a deal and pay me rent for this or pay me some fee for it and I'll do it.' Or in the case of some of the other peers, as I said earlier, they're building a building, putting equipment in it, and saying, 'Hey, rent my GPU capacity.' And that's another form of business. But I think most of the miners have been doing convertible bonds as a way to raise capital to do this, at least the larger public ones. I haven't seen a lot of Bitcoin selling from the big Bitcoin miners because most of them already sell their production volumes and I don't think we're seeing decreases in treasury holdings of Bitcoin amongst the big miners at this point. That being said, there is a huge amount of Bitcoin mining growth going on in China. China is back now as number three in the world. You're also seeing a lot of growth in places like Pakistan is coming online and Kazakhstan still remains number two in the world. So I think you're going to continue to see growth in some parts of the world. Certainly Tether has stated they're going to be the biggest Bitcoin miner in the world. There's a reason why they're doing it, as opposed to converting all those sites to AI, because it still makes sense as a business and we believe it certainly still makes sense. That being said, we are very focused on moving towards a world where inference AI is a way to better utilize certain electrons.
Okay. This statement, can you just evaluate whether or not it's true? Apparently, Bitcoin mining profitability plunged to record lows in late 2025. I don't know how does that make sense? I mean the halving cycle was last year and Bitcoin is having... You have to look at it this way. There are two things that impact a miner. One is how many Bitcoin do you get for every block that you win, and then the number of blocks that you win is directly dependent on how much hash rate you have running and what percentage of the global hash rate that makes up. So you've had global hash rate continue to grow all this year. And so it has gotten harder and harder. The difficulty rate of mining Bitcoin has gotten harder and harder. So if you haven't added capacity, you've actually gone backwards in the number of Bitcoin you're winning on a regular basis. So if your cost to mine Bitcoin, let's just say I think the industry average cost to mine Bitcoin is somewhere in the $60,000 range probably. As long as Bitcoin is trading at over $100,000, you're making good money and you're able to cover your SG&A above your variable cost nicely. But as Bitcoin comes down into the $80s or $90s, it can get a little tight for especially companies that have very expensive energy or are operating pure hosted models. If you own the data center and you're operating it yourself, your operating cost on a per kilowatt hour basis could be anywhere from 15 to 20% lower than if you're hosting with third parties. And if you look at people like American Bitcoin, they host with third parties. If you're owning and operating sites, your cost per kilowatt hour is considerably less. And at our owned and operated sites, we have some of the lowest costs in the industry.
All right. Let's talk about your diversifying strategy overall. So I'll read a statement from your 2024 December financial statement. This is the MD&A. 'In 2023, we began evaluating Kaspa as a potential way to diversify our revenue while continuing to utilize our current infrastructure and expertise in digital asset compute. At December 31, 2024, we held approximately $4.3 million or 34 million Kaspa coins on our consolidated balance sheets.' Can you comment on that development? I know it's been almost a year since December 2024 and overall your diversification strategy.
Yeah, I think the diversification strategy now is much more focused on AI and providing private cloud compute. Much less so on diversifying amongst altcoins. I mean, we still mine some altcoins because it makes sense to use some excess power to do that. There's certainly some interesting profitability in some coins. We don't detail specifically which coins and how much because it's an immaterial portion of our revenue. But I think the predominant part of our diversification is now really around inference AI, private cloud, and what we're doing to bring those types of services to enterprises around the world.
A few highlights from your mining updates in the last quarter of 2025. In September, we produced 218 blocks, a 5% increase over August. Energized hash rate increased 1% month over month. At our Texas wind farm, all containers and miners now fully deployed and connected, keeping us on track to achieve full operational status in the fourth quarter. And I believe Texas has interesting wind power and grid power updates. Can you just comment on that?
Well, I think in particular that's a site, a wind farm that we own, which we operate behind the meter at. The great thing about operating behind the meter at a wind farm that you own is that your marginal cost of energy is near zero. Your costs to operate and maintain that site are most probably somewhere around a penny per kilowatt hour. And so it's essentially free energy when the wind blows. And so we have a program we call AARP. It's a play on the statement of ARP meaning retirement. We take miners that are fully depreciated from other sites. We move them to a location like this where the power cost is near zero and we essentially have no depreciation expense on the miner because they've been paid off. So if the miners only run 50, 60, 70% of the time, we're making great money while the miners are running and they run until they eventually expire. And so this is part of our near-zero cost energy strategy. We also do this with flare gas in oil fields where we operate both in Texas and North Dakota data centers that sit in the middle of oil fields where we generate energy off of flare gas. Again, very low cost. But what's interesting is we have, as was announced contemporaneous with our Q3 earnings call, we announced a partnership we're in the process of finalizing with MPLX. MPLX is a very large mid-stream gas company and we are going to build three sites attached to their gas pipeline that will initially generate about 400 megawatts of power, scaling up to potentially 1.5 GW of power, where we will also be building three data center campuses. And that is a very large amount of energy all behind the meter, which we intend to use for optimizing mix between Bitcoin and AI operations. And we're very excited about that project. Obviously it takes time to build a power plant, but we think that we're going to make a lot of great progress on this next year and in 2027 we hope to be able to start turning on some power from those sites.
Well, with all that said, Fred, what is the investment thesis for institutional investors when it comes to Bitcoin miners like yourself? Let's just use your company for example. So Norges Bank, for example, recently took a position in Mara rated at approximately $60 million as of the writing of this particular article. Now I understand that perhaps before the Bitcoin ETF was launched last year, institutions needed to use several proxies including Mara as a proxy to be exposed to Bitcoin. But now with ETFs and perhaps other instruments available in the market, that no longer is a requirement. So I'm wondering how the narrative for institutions has shifted. Clearly there's still a tremendous amount of institutional interest.
