MARA Holdings Just Ended Its HODL Policy –  53,822 BTC Can Now Be Sold

MARA Holdings disclosed in its annual 10-K filing on March 2, 2026 that it has updated its treasury policy to allow Bitcoin sales from its entire balance sheet, ending a strategy that previously only permitted selling newly mined coins to cover operating costs.

What Changed and Why

The old policy was simple. Mine Bitcoin, sell what you need to cover costs, hold the rest. MARA’s 53,822 BTC position, valued at approximately $4.7 billion at end-of-2025 prices, accumulated under that framework over several years.

The new policy gives management the ability to sell from the full stockpile for any strategic purpose. That’s a meaningful expansion of operational flexibility, and the 10-K filing gives a clear picture of why the company needed it.

The most pressing issue is the LTV problem. MARA held $350 million in Bitcoin-backed debt at the end of 2025. As Bitcoin’s price declined toward $68,000 in early 2026, the loan-to-value ratio on that debt climbed to approximately 86.7%. Institutional lending thresholds typically sit at 70% to 75%. At 86.7%, MARA is operating with significantly less cushion than lenders prefer before triggering margin calls or forced collateral adjustments. A treasury policy that doesn’t allow selling Bitcoin creates an impossible situation when the collateral is under pressure: you need liquidity but your policy won’t let you generate it from your largest asset.

The Bitcoin lending segment added another layer of pressure. MARA earned $32.1 million in interest from lending 9,377 BTC in 2025. The problem is that trading losses and Bitcoin fair value declines in the same segment produced an $86.3 million net loss. The lending program generated income and still produced a large negative result. That outcome, combined with the LTV exposure, explains why management needed more flexibility rather than less.

The AI Pivot

The treasury policy change isn’t purely defensive. MARA is also deploying capital aggressively into AI and high-performance computing infrastructure, and Bitcoin sales are a potential funding source.

In January 2026, the company acquired a Nebraska data center for $25 million. In February, it acquired a controlling interest in Exaion for $174.5 million. It has also partnered with Starwood Capital to develop AI data centers, a partnership that requires significant cash capital. These are not small investments. The Exaion acquisition alone is larger than MARA’s annual Bitcoin lending interest income.

Riot Platforms is running the same playbook, covered earlier this week. Its AMD data center partnership and Corsicana construction plans are funded through equity issuance rather than Bitcoin sales, but the strategic logic is identical: power assets built for Bitcoin mining are being repurposed for AI infrastructure demand. The difference is that MARA now has the option to fund that transition by selling Bitcoin rather than diluting shareholders.

The LTV Math

The 86.7% LTV figure deserves attention beyond the policy change story. Bitcoin at $68,000 on $350 million in debt implies approximately $403 million in Bitcoin collateral securing the loans at the time those calculations were made. At $87,498 end-of-2025 price, the same collateral position would have produced an LTV around 67%, just within the comfortable range.

The $20,000 decline in Bitcoin price from the end of 2025 to early 2026 took MARA from a manageable LTV to one pressing against lender limits. That price sensitivity to debt structures is the same dynamic playing out across the corporate Bitcoin holder universe. It is why the current price level matters beyond simple profit and loss calculations for companies that have borrowed against their holdings.

The post MARA Holdings Just Ended Its HODL Policy –  53,822 BTC Can Now Be Sold appeared first on ETHNews.