MARA’s Bitcoin treasury decline deepened across the past year, with the company reporting 35,577 BTC at 30 June 2026, down 29% from 49,951 BTC twelve months earlier, as a $611.3 million net loss (negative $1.60 per diluted share) underscored how expensive its infrastructure pivot has become, according to the company’s MARA 10-Q filing on SEC EDGAR.
The year-over-year swing is stark: Q2 2025 produced net income of $808.2 million ($1.84 per diluted share) and adjusted EBITDA of positive $1.2 billion. Q2 2026 delivered adjusted EBITDA of negative $360.9 million, per the MARA Q2 2026 Shareholder Letter.
How the Bitcoin Treasury Decline Actually Happened
The MARA Bitcoin treasury decline is primarily a Q1 story. The company sold 20,880 BTC in Q1 2026 for roughly $1.5 billion to fund operations, repurchase debt, and seed new infrastructure investments, reducing the position from 53,822 BTC at end-2025. Q2 sales were far more modest: 2,213 BTC at an average of $73,078, offset by production of 2,422 BTC, leaving holdings at 35,577 BTC, slightly above the 35,303 BTC recorded at 31 March.
At 30 June, the structure of the treasury matters as much as the headline count. Of the 35,577 BTC: 26,307 were unrestricted, 4,742 were loaned out, and 4,528 were pledged as collateral. Post-quarter, MARA pledged an additional 18,750 BTC as initial collateral against two new Bitcoin-backed credit facilities providing $600 million of borrowing capacity. The freely deployable portion of the treasury is shrinking fast.
The 4,742 BTC on loan generated $10,651 thousand in interest income during Q2 2026, per the SEC filing, down from $13,125 thousand in Q2 2025. That is modest relative to the scale of balance-sheet Bitcoin being committed elsewhere.
Mining Output Held Up; Pricing Did Not
Operationally, the quarter was not a disaster. Energised hashrate reached 70.3 EH/s, up 22% year on year from 57.4 EH/s, and network reward share improved to 5.9% from 5.5% in Q1 2026. Bitcoin production rose 3% to 2,422 BTC.
Revenue still fell to $174.9 million from $238.5 million in Q2 2025. The shareholder letter breaks down the $63.6 million shortfall: a $65.9 million drag from lower average Bitcoin prices was only partially offset by a $7.2 million contribution from higher production. Average price of Bitcoin mined came in at roughly $71,325, against $98,975 in Q2 2025.
Energy costs moved in the wrong direction too. Purchased energy cost per Bitcoin at owned sites rose to $38,690 in Q2 2026 from $33,735 a year earlier, according to Yahoo Finance’s earnings analysis. Higher hashrate alone does not cure compressed margins when the coin price drops and per-unit energy costs climb.
The $611.3 million net loss also absorbed roughly $343 million of fair-value losses on digital assets and related receivables, the mirror image of the fair-value gains that inflated Q2 2025 earnings.
The Long Ridge Bet and What It Costs
MARA’s forward strategy centres on a proposed ~$1.5 billion acquisition of Long Ridge Energy, signed 29 April 2026 with Ohio River Partners Holdco LLC and Ohio River Partners Finance LLC, with FTAI Infrastructure as a signatory for limited purposes. FTAI’s own IR page prices the deal at approximately $1.52 billion; MARA’s filings use ~$1.5 billion. The transaction includes assumption of at least $785 million of debt, backstopped by a Barclays bridge loan, per the MARA Long Ridge 8-K.
The Long Ridge asset sits in Hannibal, Ohio: a power plant with current nameplate capacity of 485 MW, expected to reach 505 MW in H2 2026, on over 1,600 contiguous acres of industrially permitted land with water and fibre access. MARA already co-locates a data centre at the site. Combined with a 1,200-acre Texas project targeting up to 2 GW of grid capacity, management projects a potential power portfolio of about 4.8 GW.
Closing requires Hart-Scott-Rodino Act clearance and Federal Energy Regulatory Commission (FERC) approval, with the deal expected to land in H2 2026, per the MARA IR press release. Yahoo Finance reports MARA has already secured key bondholder consents, clearing one financing hurdle.
The binary for the next two quarters is whether Long Ridge closes on schedule and whether Bitcoin prices recover enough to stop the treasury from being progressively encumbered. Right now, the coin holdings that once looked like a balance-sheet cushion are increasingly functioning as collateral for the transition itself.
