MARA Holdings: An Intriguing Combination Of Bitcoin Treasury And Efficient Mining

Summary MARA first seems like a lite version of Strategy (MSTR)… because MARA also buys bitcoin to hold rather than trade. But that view shortchanges MARA. That said, there is a value-added MSTR analogy here. I like MSTR because of how its “intelligent leverage” adds value. MARA’s value-add is very different; it’s a big-time bitcoin miner. To many (including me before I looked deeply into MARA), bitcoin mining sounds obscure, exotic, and even a bit scary (think monstrous energy costs). Upon further review, and mastery of some new jargon, mining actually makes sense to me. And so, too, does MARA’s vertical integration and energy focused approach to it. MARA stock trades at a slight 2.73% discount to the market value of its bitcoin holdings. Adding-in the benefits of its approach to mining, I consider the stock a “Buy.”. Sometimes I scare myself. Some say I do that every morning when I first look in the mirror. I won’t argue. But that’s not what I mean now. I think, and I say this with great trepidation… I believe I’m starting to… drum roll… understand bitcoin! I’m sure there’s a segment of the Seeking Alpha community that reads this and yawns. “Well duh…” they (or you) may think. “What took you so long!” I don’t have an easy or self-flattering answer. All I can say is that some of us traditionalists catch on slowly. I started in the market in 1979 working for Value Line. Its founder, the late Arnold Bernhard, began his investment career in the 1920s. I’ve long been attached to value investing. I brought Berkshire Hathaway ( BRK.B ) into Value Line coverage in the mid-1990s. I’m not sure if Warren Buffett was blowing smoke or not, but he did tell me I was the first analyst to actually start covering the company regularly. Even after having moved on from Value Line, I retained enough of my fundamental training to have recognized that the many turn-of-the century dot-com high-fliers were cash-burning dregs. So I sat out that boom… and bust . In fact, in 2000 and 2002, I published books very tethered to basic fundamentals. These were, respectively, Screening the Market and The Value Connection . But underneath, I always had an affinity for Wall Street’s wild side. For a while, I covered gold-and silver-mining stocks. I even got a trip to a gold mine in Canada’s Northwest Territories near the Arctic Circle. And in the mid-1980s, I launched and managed a Value Line high yield (“junk”) bond mutual fund. (We called it “Aggressive Income.”) Now, after having retired from corporate life and working for myself rather than chasing down somebody else’s strategies and agendas, I find my taste for adventure returning. Many of the stocks you’ll find in the Analysis portion of my Profile may strike you as 180-degrees apart from Buffett-world and what you’d expect form a guy who wrote books about fundamentals. And then, on 12/20/24 , I wrote up Strategy ( MSTR ). That company is about buying and holding bitcoin. I recognizing that bitcoin is an asset. It has a price. MSTR has a quantity of it. So I could compute an asset value. Value investors love doing that! But I needed two more steps before I said “Hold” rather than “Sell.’ First, I had to understand MSTR’s “intelligent leverage” strategy. I needed that to understand why MSTR was more than just a guess on bitcoin’s next pricing move. And I had to figure out why bitcoin is not really, as Buffett famously said , “rat poison squared.” I needed to understand why sane persons would want to own it. By the time I wrote up MSTR again on 5/9/25, felt comfortable enough to say “Buy.” MARA Holdings (NASDAQ: MARA) , a partially MSTR-like company, brings my crypto self-education to the next level. I had to study up on bitcoin mining. That’s as important to MARA as intelligent leverage is to MSTR. Mining is why MARA is, like MSTR, more than just another play on bitcoin price gyrations. In One Sense, its Bitcoin “Full HODL” Strategy, We Can Think of MARA as MSTR-Lite MSTR speaks in terms of a bitcoin treasury strategy. MARA uses a different label for it… “Full HODL .” The odd acronym originated from a 2013 post in a bitcoin forum. Specifically, it was a typo committed by a user who meant to type HOLD. Since then, HODL got its own informal meaning… “Hold On for Dear Life.” As of 3/31/25, the end of the first quarter, MARA owned 47,531 coins. As of this writing, Bitcoin last traded at about $109,620 per coin. MARA’s Q1 weighted average number of diluted shares was 344.1 million. That values MARA’s holdings at $5.15 billion. Let’s now hypothetically assume MARA added no more coins since 3/31/25. In such a case, MARA’s bitcoin holdings would be worth nearly $15.14 per share. The means MARA, last (as of this writing) priced at about $14.73 per share, trades at a 2.73% discount to its bitcoin ownership. (The true discount is probably more since the company likely obtained more coins since 3/31/24.) My 5/9/25 MSTR writeup put that stock price at an approximately 116% premium to its bitcoin holdings. So, we can start out by suggesting that MARA, at a 2.73% discount, is very attractively priced relative to MSTR. But as discussed on 5/9/25, there was a lot more to the MSTR story (its intelligent leverage strategy). There’s a different sort of play with MARA. It lends out some of its bitcoin to others who sell short… and earns fee income for doing so. But there’s another much bigger aspect of the story here… MARA Doesn’t Just Own Bitcoin – It’s Also a Major Bitcoin Miner MARA separates from MSTR based on how each company gets its bitcoin. MSTR buys using its intelligent leverage financing protocol. MARA buys too. It also finances purchases. But it’s capital-raising approach isn’t as systemic as that of MSTR (at least not yet). The big differentiator is that MARA also mines its own bitcoin. I’m going to offer a little mining primer here. Just saying this company mines bitcoin, and even providing some numbers, won’t fully support an investment case for MARA. I’ll need to discuss what specialness (equivalent to MSTR’s intelligent leverage) MARA brings to this table making it more (better) than a one-to-one bitcoin play. A rudimentary understanding of mining will help you appreciate that…. Bitcoin and Mining Simplified (i.e., For People Who Still Carry Paper Currency) In the beginning (actually, on 1/3/09), Satoshi Nakamoto created bitcoin . More specifically, he launched the blockchain by creating Block 0. This is also known as the Genesis Block. He created the Genesis Block by solving a complex cryptographic puzzle. The process by which he did this is known as Proof-of-Work (PoW). Nakamoto expressed his solution a unique 64-character alphanumeric identifier (also known as a “hash”)… 000000000019d6689c085ae165831e934ff763ae46a2a6c172b3f1b60a8ce26f He also embedded a message: The Times 03/Jan/2009 Chancellor on brink of second bailout for banks (That supposedly showed Nakamoto’s motive for creating bitcoin... lack of trust in the monetary establishment). The Genesis Block awarded 50 coins (the first ones ever) to Nakamoto. And with that, bitcoin was off and running. The 50 Genesis coins can’t be spent. But new coins created by others can be and are spent, transferred, etc. Information about each new bitcoin transaction is added to something called a block. Then there’s another transaction that’s also added to the same block. Then, there’s another, and another, and another… until the block fills to capacity. (This happens on average, once every 10 minutes.) The full block closes when it fills up. And then, the race is on… Modern Bitcoin Miners in Action – Lots of Math but Only One “Eureka” Moment Bitcoin “miners” compete to add this new block to the existing “blockchain.” The latter is the official bitcoin public ledger. Miners get busy at their “rigs.” Those are super powerful computers that use ASICs (Application Specific Integrated Circuits). They’re optimized for use as mining rigs. (When connected to the bitcoin network, they’re called nodes.) Gold miners used muscle-power to find and extract gold. Bitcoin miners use brainpower to solve an insanely complex mathematical puzzle. A solution, once they come up with one, is called a “nonce.” Next, the nonce is run through a block function that leads to a “hash.” Think of a hash as the modern equivalent of a gold prospector’s choice of a spot in which to prospect. A gold miner wins by seeing gold in the rocks he brings up or in water he sifts. A bitcoin miner with a fresh new hash gets to shout “Eureka” if the hash comes in lower than a target set by the bitcoin network. Most hashes fail. They don’t meet the network’s test. To get one that works takes a lot of trial and error… and I mean A LOT ! (Shades of 19 th century gold prospectors… and then some.) With all the miners producing and trying a lot of hash solutions, here’s the challenge. In contrast to gold, there can be only one bitcoin mining winner. But suppose two, or 10, or 3,486 miners all create valid hashes? The first one to prove the validity of its solution (Proof of Work) is the sole winner. As to the other one, or nine, or 3,485 miners… tough luck, they lose… They have to try again with a new puzzle relating to the next block created by new transactions. Running Harder and Harder and Needing More and More Sustenance Are you out of breath yet? I am… just thinking about this. Imagine how much power bitcoin rigs need! That’s why these rigs, with their powerful ASIC innards, use a monstrous amount of electricity. As time passes and more coins are created, the final hash-matching challenges get increasingly harder. And the coins per win decrease due to halving events… These events keep bitcoin from reaching its hard-coded maximum circulation (21 million coins) for a long time. There are now about 19,868,488 coins . (If you click the link and see a different number, don’t tell me I messed up. The actual number changes every 10 minutes.) On day one, each winning miner got 50 new coins. Then, after issuance of every 210,000 coins, the reward per win is halved. The last such halving event happened on 4/9/24. The victory prize shrank from 6.25 coins per win to 3.125. The next halving event will cut the reward to 1.5625 coins. The upshot: We’re not going to max out tomorrow. Bitbo.io estimates that’ll happen at some time in 2040. (It estimates that by then, the reward per win will be 0.000000011641532 coins! Obviously the reward-to-effort ratio is shrinking. So expect only the fittest among the approximately 1 million unique miners to survive long term. And expect those increasingly fit miners to keep pushing their rigs harder, and for the rigs to inhale more power. MARA is The Biggest Public Bitcoin Miner Here I go again… I already bombarded you with jargon like nonce, rig, node and hash. And now, I’ll need to give you a few more terms. Hashrate tells us how much computational power a miner brings to the puzzle-solving competition. Specifically ... Hashrate is a measure of the computational power per second used when mining. More simply, it is the speed of mining. It is measured in units of hash/second, meaning how many calculations per second can be performed. Machines with a high hash power are highly efficient and can process a lot of data in a single second. Exahash means 1 quintillion hashes per second. That’s 1 followed by 18 zeros! (I’ll be you thought 1 trillion was a big number. Welcome to the bitcoin world!) We can compare miners based on EH/s… Exahash per second. Here is how MARA fares relative to the other public miners. Analyst compilation based on data from stocktitan.net and cointelegraph.com (EH/s data is the latest reported for each company on stocktitan.net , specifically here for MARA, here for CleanSpark ( CLSK ), here for IREN ( IREN ), and here for Riot ( RIOT ). All numbers of bitcoins produced comes from cointelegraph.com ) As I said MARA is the biggest. MARA is a Low Cost Miner Aside from buying bitcoins in the open market and periodically raising capital to finance its transactions, the forgoing describes what MARA does for a day-to-day living. It’s hard. So, why on earth should MARA bother mining? Couldn’t MARA just as easily buy more coins, and ask its CFO to work harder to keep finding ways to pay for them? In fact, mined coins are only small part of MARA’s total portfolio. The image below shows, for MARA, trends in blocks won and coins produced, relative to total coins owned (the holding balance). MARA IR Actually, there is a huge benefit to mining coins… despite all the virtual huffing and puffing. Typical of business in general, making something on one’s own (assuming any tolerable level of competence) will be cheaper than buying it at retail. You get to eliminate manufacturer’s margins, distribution costs, retailers’ margins. You would, of course, have your own production and capital costs. But at least you would have the power to improve your outcomes… You can manage your processes and costs. This is especially critical given halving and increases in network difficulty. This is a big deal, not just for MARA but for everybody. You see, the race to submit the winning hash differs from other races. In most contests, the finish line is fixed. A 100-yard dash is always 100 yards. That’s not so for bitcoin. Here, a 100-yard dash can turn into a 105-yard dash, and then a 110-year dash, etc. This happens because the bitcoin network adjusts the difficulty of the solution target. It does this after every 2,016 blocks (about 2 weeks). The idea here is to keep the number of winning blocks more or less constant even as more miners keep joining the network. (This is why the above mining trends chart shows, for MARA, fewer blocks won in the latest quarter.) This is why every miner, to succeed over time, must continually strive to become more operationally efficient. MARA started with an asset-lite mining model. It owned its rigs. But it paid others for hosting and infrastructure. At the peak of this phase of MARA’s history, it owned only 3% of its mining sites. That minimized capital outlays. It helped MARA achieve strong early growth in mining. Now, however, cost control is becoming more important. That’s why in 2023, MARA changed its strategy. MARA’s Pivot from its Traditional Asset Lite Approach to Pursuit of Vertical Integration The idea, here, is for MARA to own the entire bitcoin-producing chain. MARA IR Toward that end, the company made three acquisitions in 2024. These brought seven mining sites under MARA’s fold. MARA IR As a result, MARA now operates 617 Megawatts (MW) at fully-owned or controlled mining sites. That’s 55% of the 1,118 MW (1.1 Gigawatts) it now uses. The rest (501 MW) remains at hosted sites. (The latter refers to sites MARA leases from others.) MARA IR Note, however, the right column of the table… If all sites become fully optimized to their respective maximum capacities, MARA would wind up with 1.7 GW of power, and own about 70% of that total. This is already enabling MARA to slash costs. The following table shows trends in MARA’s mining cost per petahash per day. (I’m sure I made your head spin when I discussed above exahash based on 1 quintillion hashes, 1 followed by 18 zeros. Now, to stay in line with how MARA reports data, I have to introduce petahash… based on 1 quadrillion, 1 followed by 15 zeros.) Analyst compilation based on MARA data Don’t worry about variation on zeros and “…illion.” I added the percent change columns to illustrate what’s really important… the trends. Q125 is where we really start seeing what vertical integration does to costs. MARA decided to vertically integrate in late 2023. It made its acquisitions at various times during 2024. So Q125 is the first to reflect the benefits of all of last year’s purchases. Electricity Generation Looms Large in MARA’s Plans to Benefit Even More from Vertical Integration Electricity is, perhaps, THE major cost for bitcoin mining. Digioconomis.com estimates electricity cost as being 48.9% of miner income. Over the course of 2024, MARA came in well below that at its owned sites. Page 36 of its latest 10-K tells us electricity… represented 40.8%, as a percentage of our owned mining revenues for the year ended December 31, 2024. This excludes energy costs from third party hosted sites…. The average price for direct energy we paid in our owned facilities for the year ended December 31, 2024 was $0.04 per KWh. That $.04 figure refers to the 139 MW MARA-owned capacity. MARA IR MARA IR In contrast, industrial-use electricity per KWh cost averaged $0.0815 in 2024 . Helping MARA achieve its cost advantage are its behind-the-grid mining locations. At these sites, MARA can buy power directly from the source-producer. It thusly bypasses the utility ecosystem. MARA generally locates near renewable power generators, like wind farms. And looking ahead, there’s… Considerable Room for MARA to Benefit Even More from Electricity Production The idea is to capture and use what CEO Fred Thiel refers to as underutilized or stranded electrons. A lot of this relates to behind-the-grid mining sites. Often, much generated power goes to waste – is abandoned – because it can’t all be economically transported to other places, e.g., the main grid. So MARA can get this very cheaply. Besides wind, flared gas is another example. This refers to natural gas that is burned off (flared) at oil-gas production site. MARA locates modular data centers at wellheads. These are powered by gas that would be otherwise wastefully flared. MARA IR In a different sort of cooperative arrangement, MARA is a source rather than a mere user of energy. For example, in a pilot project in Finland , MARA transfers excess heat from its mining operations to a district heating system. The latter uses it to heat buildings in the community. MARA IR Meanwhile, MARA is locating in places that foster cooperation with the grid to foster load balancing . For as much of an electricity hog as mining is, it can be shut down or ramped up almost instantaneously. That opens opportunities for MARA to work with the grid to power down when it encounters peak loads. And of, course, MARA turns it up when the grid can comfortably accommodate mining. It’s now time to add another well-known power-hog, AI, to the discussion. MARA is looking to work with AI and other types of HPC (High Performance Computing). It plans to use its mining data centers to co-host for such operations. Don’t assume MARA’s data centers will have to become mega-facilities filled with the latest and greatest chips — GPUs — from Nvidia ( NVDA ). Those are most badly needed for AI training… building the models. A use of a model, (an inference event), requires much less power than a training event. Inference power demand comes from volume… many more events. The doesn’t detract from the amount of power inference needs. It’s still a lot. But there’s no need to concentrate a lot of it in a super-duper hyperscaler-friendly data center. That means AI companies can do inference from locations such as the ones MARA uses to mine bitcoin. Are visions of a new MARA revenue (hosting for AI inference) stream dancing in you mind? They should be! They are for MARA . All this should help make this stock much more than a mere speculation on bitcoin prices. (And if you’re a VERY forward thinker, you can get a sense of how MARA’s infrastructure can continue making money after bitcoin maxes out around 2040!) Risk Obviously, the future price of bitcoin is a key source of risk. If you are among those who think it and crypto in general are nonsense, you should not be in this stock. (For reasons I discussed in my 5/9/25 MSTR writeup, I disagree with that view.) Financially speaking, most of the data tables I usually show won’t be helpful. Acquisitions in pursuit of vertical integration cause unconventional revenue comparisons. And you’d likely get a mega-migraine evaluating EPS comparisons. That’s because recently-revised accounting standards let bitcoin holders record holdings at the latest market value of the coins. In the past, miners had to write them down when bitcoin fell. But they couldn’t write them up when the price recovered. That was a big headache for EPS-focused analysis. The new rule still requires holders to write down when coin prices fall. But now, they also get to write them back up when prices recover. This does better reflect the economic value of the companies. But for EPS fans, this is “OY” raised exponentially! Here is a version of my usual Financial Strength Highlights table modified to help you get a sense of MARA’s financial viability. Analyst Compilation based on Data from Seeking Alpha Financial Presentations I’ve already explained the soaring revenue. I’ve adjusted the operating margin to exclude the impact of non-cash employee stock compensation costs. (If you can’t handle dilution from employee stock options, MARA and many others among today’s growth companies aren’t for you. Employees have to be incentivized and for many such firms, cash isn’t the best way to do that right now.) I also excluded the impact of changes in the value of digital assets. These are income statement items that reflect the write-downs and write-ups I discussed based on changes in the value of bitcoin holdings. The main takeaway here is that although debt and equity soared recently (due to vertical integration), MARA is clearly solvent. I do want to mention another unusual risk source. You can see it in the recently diminished gross margin. Remember I spoke above about bitcoin halving events and increasing puzzle difficulties. As noted, these make MARA (and miners in general) work harder for rewards that shrink in relation to work done. Think of it as a business hard-wired for margins to decline. This isn’t the end of the world for MARA. It’s a fact of life. And I discussed how MARA is coping… vertical integration, cost control, digital infrastructure endeavors. These are sound moves for this business. Think of them as analogous to a company facing hard-wired industry margin erosion switching to volume-boosting strategies and cost cuts. So the halving events, etc. don’t faze me. But they might jolt investors who aren’t familiar with the dynamics of bitcoin mining… those who just look at comparisons and don’t understand what’s really going on. What to do About MARA Stock Let’s look at the price chart I usually show. StockCharts.com Frankly, I'm having trouble relating the price trends and the indicators I typically monitor (the 10- and 50-day exponential moving averages, Chaikin Money Flow, and the Chaikin Oscillator (CO), to what I see for MARA's fundamentals. Then, I decided to compare MARA prices to trends in bitcoin. Seeking Alpha Bingo! Now I get it. MARA had previously moved more or less in line with bitcoin. But that changed as 2024 progressed. MARA started to underperform bitcoin. And we know what happened in 2024... MARA changed its strategy from asset-lite to vertical integration. And that entailed huge changes in its capital structure. Add in the bitcoin halving event and the generally increasing puzzle difficulty levels. As suggested in the Risk section above, it looks to me like the Street still hasn't yet fully grasped what MARA today is truly about. It should over the long term outperform bitcoin! That’s interesting. It looks like the Street is not yet seeing all relevant aspects of MARA. That – i.e., that investors are missing important issues – is a classic example of what value investors hope to find. The value ideal is a stock with a low P/E because they don’t get it . We don’t often see that any more, in today’s information-intense world. But we do for MARA. (Maybe this investing adventure is closer to my fundamental value roots than I first realized.) I understand that nobody would actually call MARA a value stock. But can I interest anybody in creating a new word… cryptoval! Regardless… When it comes to rating stocks, I think in terms of probable future stock performance relative to the market, rather than literally buying, holding or selling. And since I’m not rating based on a quant model, I’ll eschew “Strong …” extremes. Absent an exceptional degree of conviction, I think “Buy,” Hold,” or “Sell” are enough. Over time, as the Street better understands MARA and its relationship to bitcoin prices, I see the stock as likely outperforming the broader market. I translate that to the Seeking Alpha rating taxonomy by rating MARA as a “Buy.”