MARA Shifts Bitcoin Treasury Policy, Opens Door to Selling Stockpiled Holdings
Less than two years ago, MARA’s CEO Fred Thiel was telling anyone who would listen that Bitcoin was “the world’s best treasury reserve asset” and that governments and corporations should all hold it. In July 2024, MARA announced a full HODL policy, retaining every coin mined and making open-market purchases on top of that.
The company raised over $2 billion in convertible notes that year and used the proceeds to buy more than 22,000 BTC on the open market. The conviction appeared total. However now, that conviction has quietly softened.
In a 10-K filing with the SEC on Monday, MARA disclosed that it had expanded its 2026 treasury policy to allow sales of Bitcoin already held on its balance sheet, a step beyond the partial shift it made in the second half of 2025, which had only permitted selling newly mined coins.
The company’s stock climbed 5.8% to $9.45 in premarket trading on the news, which may say something about what investors actually think of the old HODL-everything approach.
What the Filing Actually Says, and What It Doesn’t
A statement in the filing read: “In the second half of 2025, we began selling bitcoin to fund operations. In 2026, we expect to continue to monetize bitcoin opportunistically to enhance our financial flexibility,” adding that sales would be subject to market conditions and capital allocation priorities.
The word “opportunistically” is doing a lot of work there. It implies distress. Regardless, whether that framing holds up depends on context. MARA’s 53,822 BTC holdings included 5,938 BTC pledged as collateral against $350 million in outstanding credit facilities.
At year-end, when Bitcoin was priced around $87,500, that collateral was worth roughly $519 million, a loan-to-value ratio of about 67%. But Bitcoin has since slid toward $60,000. At that price, the same 5,938 BTC would be worth approximately $357 million, pushing the implied LTV to around 98%, well above the 70–75% thresholds typical of institutional margin arrangements.
MARA appears to have already moved to address this, repaying $150 million of its outstanding credit lines and entering a new $150 million facility, but the math illustrates why “opportunistic” selling and “balance sheet pressure” aren’t always easy to distinguish from the outside.
The 2025 results add to the picture. The company said it recorded a $422.2 million decrease in the fair value of its Bitcoin holdings for the year. Separately, a structured trading account funded with 2,000 BTC and managed by Two Prime, set up in Q2 2025 to pursue hedging strategies, generated a net trading loss of $22.1 million before MARA terminated the mandate in December and withdrew the remaining 1,777 BTC.
The lending segment, which generated $32.1 million in interest income from the 9,377 BTC loaned to counterparties, still recorded an $86.3 million total loss after accounting for Bitcoin’s price decline. These aren’t catastrophic numbers for a company with a multi-billion dollar Bitcoin position, but they’re not nothing either.
MARA: The AI Pivot and What It Costs
There’s a second thread running through this. MARA has been repositioning itself as something beyond a straightforward Bitcoin miner, leaning into artificial intelligence infrastructure and high-performance computing.
CEO Fred Thiel, in a February 26 press release alongside the SEC filing, said the company’s “power-rich sites give customers what they need most: predictable access to energy at scale.”
In January, MARA acquired a 42-megawatt operational data center in central Nebraska for $25 million in cash. That kind of capex needs funding, and selling Bitcoin, even at a loss relative to prior valuations, is one cleaner option than issuing more equity into a declining stock price.
MARA isn’t alone in this. Core Scientific sold roughly 1,900 BTC in January for $175 million, around $92,100 per coin, as part of its own pivot to AI infrastructure. The pattern among public miners appears to be converging: accumulate Bitcoin aggressively when prices are rising, then quietly start monetizing when margins compress, capital needs grow, and network difficulty keeps climbing regardless.
MARA’s mining output fell 7% year-over-year to 8,799 BTC in 2025, a decline attributed to the April 2024 halving and rising network difficulty even as its hashrate grew to 66.4 EH/s. Mining fewer coins while holding more debt is a combination that tends to concentrate minds around liquidity options.
For the broader market, the question is one of scale and timing. MARA controls 53,822 BTC, the second-largest corporate Bitcoin position after Strategy. How much of that gets sold, when, and into what kind of order book will matter.
A few hundred coins moving through institutional desks on a strong day barely registers. A larger, more urgent liquidation into thin weekend markets is a different story.
The filing offers no specific targets or timelines, which is perhaps the point, the optionality is the asset. Whether MARA uses it surgically or out of necessity is a question the next few quarters will answer.
