Summary MARA Holdings,Inc. (MARA) is a leading Bitcoin treasury miner and the second-largest corporate Bitcoin holder, but remains highly vulnerable to Bitcoin price volatility. MARA trades at an 96% premium to its net Bitcoin asset value, justified by its mining operations and future diversification into AI and High Performance Computing. The Bitcoin deployment strategy is not a meaningful contributor to MARA's earnings or valuation. Given these factors, MARA is fairly valued at present and warrants a HOLD rating due to its vulnerability to Bitcoin price swings amid long-term upwards trend. Editor's note: Seeking Alpha is proud to welcome Louis Rose as a new contributing analyst. You can become one too! Share your best investment idea by submitting your article for review to our editors. Get published, earn money, and unlock exclusive SA Premium access. Click here to find out more » My thesis is that MARA Holdings, Inc. (MARA), although more than a simple Bitcoin treasury company, is still too vulnerable to Bitcoin price fluctuations to deserve a higher valuation today. My arguments are that, first of all, the company should be valued as a Bitcoin treasury, a Bitcoin mining company, and a future AI / HPC company. Indeed, their Bitcoin deployment strategy does not contribute meaningfully to their financials, as we will see later, and their recent moves towards AI / HPC, although promising, are still at the proof-of-concept stage. Then, regarding Bitcoin mining, the profitability of their operations is highly dependent on Bitcoin price fluctuations, which is a real risk for the company. Moreover, their capacity for the number of Bitcoins mined in the future is already roughly priced in the stock. Although Bitcoin’s progressive adoption by institutions is a probable long-term catalyst for the stock, I don’t see a reason to attribute them a higher multiple to NAV today. I rate MARA Holdings as a “HOLD”. Not a Simple Bitcoin Treasury Company Founded in 2010, MARA Holdings (formerly Marathon Digital Holdings and Marathon Patent Group) originally started as a patent holding company. By the end of the 2010s, the company had completely shifted its line of business to cryptocurrency. Since 2020, the company has operated in Bitcoin mining at scale, being one of the largest public miners by hashrate (meaning mining power) today. In addition to mining, they also buy and hold a very large amount of Bitcoin, being today the second largest bitcoin corporate holder worldwide behind Michael Saylor’s Strategy Inc. (MSTR), formerly Microstrategy . As of November 15th, 2025, MARA owns 53,250 Bitcoins. That’s 0.265% of the 19,948,812 Bitcoins mined to date , and 0.254% of all the 21,000,000 Bitcoins that will ever be mined. Top 12 Public Bitcoin Treasury Companies (bitcointreasuries.net) At the time of writing, Bitcoin’s price of $92,900 values MARA’s Bitcoin reserve at $4.95 billion. If we subtract from that their $3.52 billion in total long-term liabilities and add back their $0.82 billion in cash and cash equivalents ( see their 10-Q here ), we obtain a $2.25 billion theoretical market capitalization, solely from their Bitcoin reserve. Their current market capitalization of $4.41 billion implies a mNAV (multiple-to-net-asset-value) of 1.96x. This means that MARA Holdings currently trades at a 96% premium compared to the net value of their Bitcoin holdings. This premium is explained by the fact that, contrary to Strategy, MARA Holdings is not a simple Bitcoin treasury company. As we have seen, they mine Bitcoin: over the last 5 quarters, around 2000 to 2500 Bitcoins per quarter. They also actively deploy their Bitcoin holdings: in loans (19.7% of all Bitcoin holdings) or as collateral for credit lines (9.6%). Finally, they are now expanding into AI/HPC (High Performance Computing). They plan on diversifying their computing power use in private cloud computing (More specifically, in edge AI inference). Their recent acquisition ( pending regulatory approval ) of a 64% stake in Paris-based startup Exaion for $168M supports this strategic shift, as Exaion has expertise in tier III/IV data center management and AI/HPC. With respect to their mining operations, they are switching from an asset-light model to a vertically integrated model, where they directly own a growing part of the full mining pipeline: the energy production facility, the data center and the Bitcoin mining equipment. Back in 2023, 97% of their capacity was hosted at third-party sites. In Q3 2025, that share was reduced to 49%: 1099 of all the 2144 Bitcoins produced during the quarter were from their own operations. As stated in their Q325 shareholder letter , this strategic shift from an asset-light model to a vertically integrated model does negatively impact their current expenses — namely, their capital expenditures and general and administrative expenditures, see their 10-Q — but improves their cash-flow generation prospects. Indeed, as MARA Holdings phases out third-party contracts over time, they expect their Bitcoin mining unitary costs to continue to decrease. Thanks to this ongoing strategic pivot, they have already reduced their cost per petahash per day by 15% between Q3 2024 and Q3 2025, which helped them reach a purchased energy cost per Bitcoin of $39.2 thousand. Their declared goal is to reduce the third-party sites' contribution to 30% by the end of 2025. To continue to make progress towards that goal, they recently entered into a partnership with MPLX LP (formerly known as Marathon Petroleum) to gain direct access to the Delaware Basin energy resources (natural gas). MARA will build and operate power plants on-site that will convert the natural gas to electricity, which will power both MARA data centers and MPLX operations. Regarding their diversification into AI and High Performance Computing, MARA deployed their first 10 AI inference racks at the Granbury, Texas site in Q3 2025 (see the Q3 earnings presentation , slide 5). This is the beginning of their effort to leverage their infrastructure for AI applications. As a 255% increase in AI power demand is forecasted between 2025 and 2030, MARA Holdings strategically positions itself at the confluence of energy, compute, cryptocurrency and AI. MARA’s Bitcoin Deployment Strategy Isn’t a Significant Contributor to Earnings MARA’s Bitcoin deployment strategy encompasses loans (19.7% of all Bitcoins), active management (3.6% of all Bitcoins) and credit collaterals (9.6% of all Bitcoins). MARA's Bitcoin deployment overview (MARA's Q3 2025 publication) First, Bitcoin’s use as a credit collateral is simply a financing activity and does not contribute to profits per se. Then, the active management line of business can also be considered non-meaningful, as only a very small portion (3.6%) of owned Bitcoins is concerned. The main contributor to their deployment strategy is the loans: in Q3 2025, MARA declared an interest income of $17.67 million. When dividing this $17.67 million interest income by the 10,377 loaned Bitcoins, this equates to an annual yield of 5% to 7% depending on Bitcoin’s market price, which is in line with standard yields for the practice. Still, MARA’s $17.67 million interest income has to be compared to their interest expenses of $12.76 million for the quarter: their net interest income is thus $4.91 million. I state that this cannot be a significant contribution to their market capitalization. Even when attributing a generous 30x P/E ratio, this would only contribute $147 million to their market capitalization, which is currently $4.53 billion. As such, I see MARA’s Bitcoin deployment strategy as a way to comfortably cover their interest expenses, but not much more than this. I conclude that this line of business should not be considered as a meaningful contributor to their market capitalization. MARA’s Diversification Into AI is Promising, but Still at the Proof-of-Concept Stage In their Q3 2025 shareholder letter, MARA Holdings states that their energy expertise is a strategic advantage in this new era of Artificial Intelligence, where growing AI demand will be constrained by the energy supply. MARA’s long-term goal is to allocate a portion of their computing power to AI inference on the edge. The day-to-day repartition of computing power between cryptocurrency mining and AI inference will depend on where the higher profitability lies. My argument is that, while having a lot of potential, this diversification into AI is still at the proof-of-concept stage today, as the numbers will show. As such, the potential of diversification into AI inference can only be partially accounted for in the stock price to date. In Q3 2025, MARA Holdings installed ten AI racks within a non-water-cooled modular data center at its Granbury, Texas site, which has a 225 MW operational capacity. For reference, MARA Holdings' total operational capacity across all sites culminates at 1.2 GW. Although MARA did not communicate the exact power capacity allocated to AI inference at the Granbury, Texas site, we can make a rough estimation for it. Today, modern GPU racks consume 10 to 15 kW per rack. This means that, currently, MARA’s AI inference operations only absorb 0.12 MW (average estimate) out of the 1200 MW available. Meaning, about 0.1% of the total available power. For MARA Holdings, this first step into AI diversification echoes the recent strategic shifts from other, more advanced Bitcoin miners: TeraWulf, backed by Alphabet Inc. (GOOGL) or Core Scientific, Inc. (CORZ). According to this VanEck 2024 report, if Bitcoin miners were to allocate 20% of their electrical capacity to AI, this could justify a doubling of the stock price. This is explained by the fact that AI workloads can generate more revenue than Bitcoin mining operations, while reducing exposure to Bitcoin's price and reducing infrastructure costs. Bitcoin miners have a time-to-market advantage: repurposing a Bitcoin mining site into an AI data center takes, on average, about a year, while building one from scratch takes 3 to 6 years. Regarding profitability potential, the Core Scientific example is revealing: they contracted a total revenue of $6.7B over 12 years with CoreWeave, for a final capacity projected at 500 MW. For Core Scientific, this entails several advantages: a reduced exposure to Bitcoin's price, a guaranteed revenue for 12 years, and having infrastructure upgrades paid by the customer. Considering that MARA Holdings is very early in its AI diversification pivot, and is late compared to a number of competitors, I only attribute the company a 30% premium for its AI diversification plans. MARA’s Bitcoin Mining Activity Alone Does Not Justify its Current Valuation Between Q3 2024 and Q3 2025, MARA mined approximately 9200 Bitcoins. Let’s analyze the profitability of their mining activity (specifically for their fully-owned sites, which should represent the vast majority of their mining activity as soon as 2026). The chart below also shows Bitcoin purchases by MARA Holdings, which are independent from their mining business. Large-scale cryptocurrency purchases are simply a bet on the coin’s future price, and fall under the bitcoin treasury part of their business. How MARA grew their Bitcoin reserves over the last 5 quarters (MARA's Q3 2025 publication) Let’s estimate the profitability of MARA’s mining business. MARA’s bitcoin production fluctuates between 2000 and 2500 Bitcoins per quarter. They are the largest publicly traded Bitcoin miner, with an energized hashrate of 60.4 EH/s. For clarification, 1 EH/s means one quintillion hash computations per second. The total network hashrate for Bitcoin is around 1100 EH/s . This means that MARA Holdings contributes about 5.5% of all computing power allocated to Bitcoin mining worldwide. For their wholly-owned sites specifically, the energy purchase cost is $39.2 thousand per Bitcoin, one of the lowest in the sector. But several other costs must be accounted for. Operating and maintenance costs, as well as general and administrative costs. Using their 10-Q numbers, operating and maintenance costs came at $12.9 thousand per Bitcoin in Q3 2025. Regarding general & administrative costs, I state that most, but not all, of those costs should be attributed to the Bitcoin mining activity. Let’s take a lenient hypothesis and allocate only 70% of general and administrative costs to their mining activity. This leads to a total cost per Bitcoin mined of $79.3 thousand. Using the current Bitcoin price of $95,750, the cash operating margin is 20.7% (including 70% of general and administrative expenses). Were we to attribute 90% of general and administrative expenses to mining, the break-even price per Bitcoin would be $88 thousand, and the operating margin would fall to 9%. As we can see, the Bitcoin mining margins are highly dependent on Bitcoin price fluctuations. If the Bitcoin price were to drop below $80K for a sustained period, this could spell trouble for the company, which would still have to finance its various expenses. To support the argument that expenses financing is a potential risk for MARA, the company recently declared that “they may now choose to sell a portion of the Bitcoin produced from mining to support their ongoing operating expenses” (see page 14 of the Q3 2025 shareholder letter). We also have to consider that Bitcoin mining tends to become more difficult over time: the number of remaining Bitcoins to mine steadily shrinks, mining equipment depreciates and becomes obsolete, total network hashrate tends to grow over time due to increasing competition, and successive halvings reduce the number of Bitcoins granted per block won. If MARA Holdings were to keep their 5.5% share of total Bitcoin mining compute over time, they would obtain about 57,800 Bitcoins out of the remaining 1,051,800 to mine. That would increase their current Bitcoin reserve by 108%. If we take into account the costs associated with mining those Bitcoins, and assume a 20% cash operating margin, MARA would make $1.07B in profit by mining those remaining 57,800 Bitcoins at current market prices. That’s a x1.47 bonus from their current Net Asset Value. When taking into account the 30% valuation premium from the AI diversification shift, I estimate that MARA Holdings currently deserves a x1.91 multiple to NAV. They are currently selling at a x1.96 multiple. I conclude that MARA Holdings' current valuation already prices in the Bitcoin mining business, as well as their diversification into AI. MARA’s Net Income Evolution Corroborates my Thesis When looking at MARA Holdings' Net Income, the overall thesis is verified: the net income is highly volatile, sometimes negative, and depends mostly on Bitcoin’s price movements. MARA Holdings Net Income over the last 5 quarters (MARA's Q3 2025 Publications) For example, in Q3 2025, MARA declared a Net Income of $123.1M. But Bitcoin’s price increase from June 30, 2025, to September 30, 2025, actually contributed $343M to their quarterly net income. If the Bitcoin market price had been flat during the quarter, MARA’s Net Income would actually have been negative. This supports the thesis that MARA Holdings' financial performance is highly reliant on Bitcoin’s price appreciation over time, and that the company does not generate significant value aside from holding Bitcoin that appreciates over time, and mining Bitcoin at a positive but highly volatile profit margin. Conclusion I conclude that MARA Holdings' current market capitalization correctly reflects the current value of their Bitcoin holdings, of their Bitcoin mining business and of their future growth potential from AI/HPC diversification. As discussed, their Bitcoin deployment strategy does not constitute a significant contribution to their earnings power to date. Although the long-term trend for Bitcoin is upwards as institutional adoption progresses worldwide, MARA Holdings' financial health is still highly dependent on short-term and medium-term Bitcoin price fluctuations. As such, I consider MARA Holdings to be fairly valued currently. I would need to see more execution on the AI diversification side, as well as a reduced exposure to Bitcoin's price, before rerating the stock. My recommendation: HOLD.
