MARA’s Reported $81M Bitcoin Transfer Is Not Proof of a Sale
Lookonchain reported that MARA Holdings moved 996 BTC, worth about $81.13 million, to a wallet it identified as linked to Galaxy Digital. Neither the wallet label nor the transfer confirms that MARA sold the bitcoin or that Galaxy bought it.
- Reported transfer: Lookonchain reported 996 BTC moved on October 9, 2026.
- Sale unconfirmed: MARA’s documented March bitcoin sale was a separate transaction.
- Holdings declined overall: MARA reported 53, 822 BTC at the end of 2025 and 35, 577 BTC at June 30, 2026.
- AI capacity is planned: MARA’s Starwood development plans cite about 1 GW of near-term computing capacity, not capacity already online.
What the transfer does and doesn’t show
On October 9, 2026, Lookonchain flagged the 996 BTC movement and identified the destination wallet as linked to Galaxy Digital. That identification does not establish that Galaxy received the bitcoin as a buyer or that a trade took place. MARA has not confirmed the reason for the transfer.
Companies may move bitcoin between their own wallets, send it to a custodian, or transfer it ahead of a potential transaction. Until a completed trade is confirmed, “transferred” is the accurate description. The movement is not evidence of another sale just because MARA has sold bitcoin before.
MARA’s March sale was a separate transaction
MARA said it sold 15, 133 BTC between March 4 and March 25, 2026, for approximately $1.1 billion. The company expected to use most of the proceeds to repurchase its 2030 and 2031 convertible notes. Any remaining funds would be available for general corporate purposes.
Convertible notes are debt that can be exchanged for company shares under specified terms. Repurchasing them can reduce debt and the potential for future share dilution. The effect depends on the notes’ terms and the company’s actions.
MARA agreed to repurchase about $1.001 billion in note principal for approximately $912.8 million in cash. The company estimated that the transactions would capture about $88.1 million in value before costs, at roughly a 9% discount to face value, and reduce outstanding convertible debt by approximately 30% if completed as expected. MARA described the buybacks as a way to strengthen its balance sheet and reduce potential dilution.
These details support the view that the March sale was part of a debt-management strategy. They do not explain why the 996 BTC moved in October.
Bitcoin holdings fell overall, but not in a straight line
StockTitan’s summaries of MARA’s reported results put the company’s bitcoin holdings at 53, 822 BTC on December 31, 2025, and 35, 577 BTC on June 30, 2026. That’s a substantial decline between the two dates, but the figures do not show a continuous drop.
MARA reported holdings of 35, 303 BTC at March 31, then 35, 577 BTC at June 30, an increase of 274 BTC. Mining, sales, purchases, and transfers can all change a company’s bitcoin balance. Snapshots alone do not explain every movement.
StockTitan’s summary of MARA’s second-quarter 2026 results reported revenue of $174.9 million, down 27% from $238.5 million a year earlier, and a net loss of $611.3 million. It attributed much of the loss to a $343 million unrealized loss on bitcoin. An unrealized loss reflects a change in an asset’s reported value. It is not the same as selling the asset at a loss, though it can still weigh heavily on reported earnings.
Data-center plans are not operating capacity
MARA is also pursuing digital-infrastructure projects beyond bitcoin mining. Its arrangement with Starwood Digital Ventures covers the development, leasing, and marketing of sites for hyperscale, enterprise, and AI-capable infrastructure. The plan cites about 1 gigawatt of near-term IT capacity, meaning the computing capacity a site is designed to support.
That figure describes planned capacity, not 1 GW of data centers already built and running. Development still depends on securing customers and power, financing projects, and completing construction. The plans could give MARA another use for its experience with energy-intensive computing, but a development pipeline is not operating revenue.
The expansion could broaden MARA’s business over time. It does not remove the company’s exposure to bitcoin-price swings or the costs and execution risks of large infrastructure projects.
Key questions and answers
-
Did MARA sell the 996 BTC?
That has not been confirmed. Lookonchain reported a transfer to a wallet it identified as linked to Galaxy Digital, but a wallet label does not prove a sale or a purchase by Galaxy.
-
How much bitcoin did MARA sell in March 2026?
MARA said it sold 15, 133 BTC between March 4 and March 25 for approximately $1.1 billion, with the proceeds intended mainly for convertible-note repurchases.
-
How much principal did MARA repurchase?
MARA agreed to repurchase about $1.001 billion in note principal for approximately $912.8 million in cash. The principal amount and cash paid are different figures.
-
Is the planned 1 GW of computing capacity already online?
No. It is planned near-term capacity tied to development work, not confirmed operating capacity.
-
Do MARA’s earlier bitcoin sales explain this transfer?
No connection has been established. The March sale is documented, but it does not confirm the purpose of the separate October movement.
MARA’s March sale shows the company has used bitcoin to manage debt while pursuing businesses beyond mining. That context makes the reported October transfer worth watching, but it does not tell us what happened to the 996 BTC. Until a sale is confirmed, “transferred” is the fact. “Sold” remains speculation.