A bitcoin mining company has started actively selling its own BTC—and sold more than 13,000 in a single transaction.
What makes this worth paying attention to is not that a new wave of bitcoin sell orders has suddenly hit the market, but that MARA Holdings is changing how it has managed its assets in the past.
According to the company’s latest 2026 Q3 shareholder letter, MARA sold 13,597 BTC after the end of the third quarter at an average price of approximately $81,039, raising about $1.1 billion.
Of this amount, $750 million was used to fully repay a bitcoin-backed loan, thereby releasing the 17,477 BTC that had previously served as collateral.
More importantly, the company said it may continue to sell some of its BTC in the future, depending on market conditions, to fund AI infrastructure investments, repay debt, and conduct other capital allocation activities such as share buybacks.
According to the company's disclosures, it sold 26,984 BTC during the first nine months and another 13,597 BTC after the end of the third quarter, bringing total sales since the start of 2026 to approximately 40,581 BTC.
For a company that started out in Bitcoin mining, this is no longer a simple routine coin sale; it is a strategic adjustment to the balance sheet.
I. Why would MARA rather sell BTC than repay its loans first?
First, let's look at where the proceeds from this transaction went.
MARA previously raised $750 million through Bitcoin-backed loans, including both new financing and the refinancing of existing loans.
This type of financing is not uncommon among mining companies.
A company can retain its BTC while obtaining cash against it for acquisitions, expansion, or other business investments.
But the problem is that using BTC as collateral is not without cost.
If the price of Bitcoin falls sharply, a company may face risks such as a decline in collateral value, margin calls, and changes to financing terms.
Even if the price of BTC rises, collateralized assets cannot be used as flexibly as assets held entirely free of encumbrances.
MARA chose to sell 13,597 BTC, raising approximately $1.1 billion, and used $750 million to repay loans.
This transaction had three direct results:
* All loans were repaid, reducing the debt burden.
* The 17,477 BTC previously pledged as collateral for loans were released.
* The share of the company's total BTC holdings that was pledged or lent fell below 5%.
This means MARA has proactively reduced its reliance on BTC-backed financing.
I think this is the most important part of the move.
In the past, many mining companies treated BTC as a long-term reserve asset and tried to minimize sales. But as a company enters a new stage of development, cash flow, debt structure, and investment returns may matter more than simply increasing the amount of BTC it holds.
II. Selling coins does not necessarily mean being bearish on Bitcoin—MARA really wants to change its business model
If you look only at the sale of 13,597 BTC, it is easy to conclude that MARA is no longer optimistic about Bitcoin's future.
But based on the strategy the company has announced, that interpretation is incomplete.
MARA is attempting to transform itself from a traditional Bitcoin mining company into an operator of AI and high-performance computing infrastructure.
There is a very practical business rationale behind this.
Bitcoin mining requires substantial electricity, land, equipment, and infrastructure. AI data centers likewise need power resources, server facilities, cooling systems, and high-density computing environments.
For mining companies that already have power resources and data center infrastructure, these assets may have new commercial uses.
Compared with relying solely on mining rewards, signing long-term computing power or infrastructure leasing contracts with AI companies could create another source of revenue.
In its latest shareholder letter, MARA said it is applying the power resources and infrastructure capabilities it has built up over many years to its AI/HPC business.
As of the third quarter, the company had a power portfolio of approximately 1.9 GW and outlined plans to expand it to 4.8 GW.
The company also advanced its Matagorda County project in Texas, whose planned site capacity could reach up to 2 GW.
This shows that MARA wants to further convert the energy and infrastructure previously used for Bitcoin mining into productive resources for the AI industry.
Of course, having power resources does not mean that it already has a mature AI cloud computing business. Project construction, customer contracts, equipment deployment, and future cash flow all need time to prove themselves.
What is certain, however, is that MARA is looking for more sources of revenue than mining alone.
III. An easily overlooked detail: MARA's profits do not come entirely from mining
There is another set of figures in this earnings report worth examining closely.
MARA generated approximately $164.2 million in revenue in the third quarter, down 35% year over year, and net income of approximately $632.1 million.
At first glance, its net income looks exceptionally strong.
But a closer breakdown reveals that it included approximately $885.3 million in gains from changes in the fair value of digital assets, primarily due to fluctuations in the price of BTC during the quarter.
This means net income cannot be equated directly with cash profits generated by the company's mining and infrastructure businesses.
Similarly, the company's reported adjusted EBITDA was also significantly affected by changes in the fair value of digital assets.
Why emphasize this point in particular?
For Bitcoin miners, rising asset prices can substantially improve reported profits, but that does not mean mining efficiency, operating cash flow, and the scale of actual business operations have grown at the same time.
When analyzing companies like this, you need to distinguish at least three things:
First, whether the mining business itself can generate sufficient operating cash flow.
Second, how much changes in the value of its BTC holdings have affected reported profits.
Third, whether its AI infrastructure business can eventually become an independent and stable source of revenue.
If you lump these three things together, it is easy to overestimate a mining company's true operating capacity.
IV. Mining companies may follow two very different paths in the future
MARA's move has made me start paying attention to the growing divide in the future of Bitcoin mining companies.
The first path is to keep Bitcoin mining at the core of the business.
These companies compete primarily on computing power, energy costs, and mining efficiency. Their performance is closely tied to the price of BTC, the network's total hashrate, and mining difficulty.
The second path is to redirect existing power and data center resources toward AI/HPC businesses.
These companies need to invest more capital in building computing infrastructure, sign service contracts with AI customers, and gradually develop new sources of revenue.
But neither model is inherently better or worse.
Mining directly involves participation in the Bitcoin network, while AI infrastructure businesses must deal with customer creditworthiness, project financing, equipment depreciation, and fulfilling long-term contracts.
MARA has chosen to pursue both paths in parallel.
It has not abandoned Bitcoin entirely. Instead, it views BTC as an asset that can be held, sold, lent, or used for capital allocation as needed.
This also means that, in the future, evaluating a mining company may require looking beyond how much BTC it holds to consider how much power it controls, how many projects it has under contract, and how much cash flow those assets can generate.
V. What can ordinary BTC holders learn from this?
I think MARA's move offers a very practical lesson for ordinary digital asset holders:
Holding an asset and managing an asset are two different things.
Many people who follow BTC focus only on whether its price is going up or down, and rarely consider when they might need to use their funds, whether they need to convert them, or how to allocate funds across different markets.
MARA's approach offers a real-world example at the corporate level:
When a company has a new investment project, it can assess the opportunity cost of holding BTC and then decide whether to sell some of its assets.
Ordinary users, of course, cannot simply copy a company's capital management approach, but they can learn from its approach to managing funds.
For example:
* If you have a clear need to spend US dollars in the future, you can check the exchange rates and fees for converting U to USD in advance.
* If you want to explore traditional technology companies, you can learn how to use U to trade actual U.S. and Hong Kong stocks.
* If you still want exposure to digital asset markets, you can distinguish between spot holdings and derivatives trading rather than putting all your funds into highly volatile trades.
* If you need to move money internationally, consider the remittance channels, arrival times, and specific requirements of the destination.
In these situations, BiyaPay is one tool worth learning about.
BiyaPay uses U as its primary funding currency, supports converting U to USD, and provides access to actual U.S. and Hong Kong stock trading. U.S. stock trading on the platform is commission-free, and eligible dividend benefits are supported.
For people who hold U and want to explore U.S. technology stocks, they can learn about the stocks available on the platform and the actual trading conditions, connecting digital asset funds with use cases in traditional stock markets.
BiyaPay also offers digital asset spot and derivatives trading, advertises zero fees and certain Maker 0-fee features, and provides commodity futures and flexible savings products. The displayed maximum annualized rate for flexible savings is 10.22%. Actual returns, redemption terms, and product risks are subject to the applicable terms.
For international fund transfers, the platform also supports cross-border remittances to 108 countries worldwide and exchanges between more than 30 fiat currencies and over 200 digital currencies. In eligible regions, local remittance channels can reduce intermediary processing steps and improve transfer speed. Actual arrival times depend on the destination and the specific channel.
I think what makes BiyaPay worth learning about is not just digital asset trading, but the way it connects different financial needs, including U conversion, global stocks, commodity futures, and cross-border remittances.
For people who follow BTC, U.S. stocks, and global assets at the same time, managing funds should involve more than deciding which assets to hold; it should also take into account how different assets can be used.
VI. Three more developments at MARA worth tracking
First, how much more BTC will it sell?
The company has stated clearly that it may continue selling some BTC in the future to fund AI infrastructure, manage debt, and support other capital allocation activities.
Therefore, the amount sold, the sale price, and how the proceeds are used will directly affect how the market views its BTC reserve strategy.
Second, when will the AI infrastructure business generate stable revenue?
MARA already has a sizable power portfolio, but turning power resources into data center leasing revenue requires project development, approvals, construction, and customer delivery.
Key indicators to watch include contract signings, project progress, and actual cash flow.
Third, will debt below $50 billion continue to decline?
More precisely, the company's board has authorized a share repurchase program of up to $500 million, while the company also needs to address future put options on its convertible notes.
Authorization for a share repurchase does not mean the buyback has already been completed. Actual execution will still depend on market prices, the company's financial position, and its capital allocation decisions.
If MARA can reduce debt while turning BTC and power assets into more stable operating revenue, its business model may undergo a more profound transformation.
Conversely, if investment in AI projects continues to grow without generating corresponding contracts and cash flow, the market may reassess the costs of this transition.
Ultimately, what I find most worth examining about MARA's BTC sale is not whether it is bullish or bearish on BTC, but that it is beginning to redefine BTC's role on the corporate balance sheet.
Bitcoin can be a long-term reserve, a financing tool, or a source of funds for building new businesses when the time is right.
For a company, the true value of an asset lies not only in how much it holds on its books, but also in whether it can create new operating value through sound allocation.
From mining to AI infrastructure, and from holding BTC to actively managing assets, MARA is attempting a business model transformation.
Whether this path will succeed remains to be proven by future results.
This article is for informational purposes only. Please do your own analysis and make your own judgments. It is not investment advice. Avoid unnecessary losses!