Corporate Bitcoin Holders Split: Miners Sell While Accumulators Double Down

The Great Corporate Bitcoin Divergence: Who's Selling, Who's Buying, and What It Means

The corporate Bitcoin treasury movement, once celebrated as a unified institutional vote of confidence in the asset, is fracturing along a fault line that reveals something deeply important: not all corporate Bitcoin holders were created equal. As market conditions tighten, the distinction between companies that hold Bitcoin as a genuine long-term reserve asset and those that accumulated it opportunistically — or without the balance sheet strength to sustain the position — is becoming impossible to ignore. What we are witnessing right now is not a crisis of confidence in Bitcoin. It is a stress test of corporate Bitcoin strategy, and the results are brutally clarifying.

On one side, distressed miners and overleveraged treasury companies are liquidating significant portions of their holdings to meet debt obligations and fund operational pivots. On the other, conviction-driven accumulators are buying into weakness with disciplined consistency. The divergence tells us more about corporate finance than it does about Bitcoin's fundamentals — but the selling pressure is real, and its market implications deserve serious scrutiny.

The Facts

The volume of Bitcoin being moved out of corporate treasuries in recent months is substantial. MARA Holdings executed the largest single corporate liquidation in this cycle, selling 15,133 BTC between March 4 and March 25 for approximately $1.1 billion in proceeds ^[2]. The company directed those funds toward repurchasing its 0.00% convertible senior notes due in 2030 and 2031, reducing outstanding debt by roughly 30% and shrinking its holdings from 53,822 BTC at the start of the year down to 38,689 BTC ^[2][3].

Riot Platforms, one of the largest publicly listed Bitcoin miners in the United States, recorded a wallet outflow of 500 BTC — worth approximately $34 million — flagged by blockchain intelligence firm Arkham on April 1 ^[1][2]. The company had not publicly commented on the transaction at time of reporting, but the movement to an exchange-linked address is consistent with a sale ^[2]. The move follows an estimated $200 million in Bitcoin sales during the final months of 2025, as Riot redirects capital toward artificial intelligence and high-performance computing infrastructure ^[2]. This comes despite Riot posting record 2025 revenue of approximately $647 million ^[1].

Smaller entities are also reducing exposure. Genius Group, an AI-focused education company that once held up to 440 BTC, has fully exited its Bitcoin position, selling its remaining 84 BTC to retire $8.5 million in debt ^[2]. Nakamoto Holdings sold 284 BTC for around $20 million in March — roughly 5% of its reserves — to fund working capital following acquisition activity, against a backdrop of a reported pre-tax loss of $52.2 million for 2025 ^[1][2]. Empery Digital transferred out what on-chain trackers described as its remaining 1,795 BTC, worth approximately $122.5 million, to Gemini following a series of smaller sales throughout March, leaving the firm holding 2,989 BTC — down from a peak of around 4,000 BTC ^[1][2]. Even the Kingdom of Bhutan has continued reducing its state-backed Bitcoin position, selling a total of 3,103 BTC, down significantly from a peak holding of over 13,000 BTC in October 2024 ^[2].

Contrasting sharply with this wave of selling, Metaplanet — the Tokyo-listed investment firm — announced on April 2 that it acquired 5,075 BTC during Q1 2026 at prices averaging between $78,000 and $79,898 per coin, bringing its total holdings to 40,177 BTC ^[3]. That accumulation pushes Metaplanet past MARA's current position, securing its rank as the third-largest publicly traded Bitcoin holder globally, behind Strategy with over 762,000 BTC and Twenty One Capital with 43,514 BTC ^[3]. CEO Simon Gerovich has framed the strategy as a long-term hedge against yen depreciation and Japanese inflation conditions ^[3]. Metaplanet funds its accumulation through equity issuances, warrant sales, and a Bitcoin options income strategy that generated approximately 2.97 billion yen in Q1 2026 revenue ^[3]. The company has set an ambitious target of 210,000 BTC by end of 2027 ^[3].

For context, public companies collectively still hold approximately 1.16 million BTC — more than 5% of Bitcoin's fixed supply of 21 million coins — meaning the asset class remains deeply embedded in corporate balance sheets despite the recent selling wave ^[2].

Analysis & Context

The pattern emerging here rhymes with dynamics seen in previous Bitcoin market cycles, but with a new institutional dimension. When Bitcoin corrects significantly from its highs — in this case with BTC trading near $66,000, roughly 32% below Metaplanet's average cost basis and well below the highs above $100,000 seen in late 2024 — the weakest hands are always the first to fold. In prior cycles, those weak hands were predominantly retail investors. Today, they are increasingly leveraged corporate vehicles that built Bitcoin positions using convertible debt, term loans, and equity dilution without sufficient operational cash flow to weather a prolonged drawdown.

The critical distinction is structural. Miners like MARA and Riot face a double pressure: the Bitcoin price decline compresses their revenue per coin mined while operating costs — energy, hardware, and debt servicing — remain largely fixed. Selling Bitcoin is not a sign of ideological retreat; it is a rational financial response to a margin squeeze. But it underscores a fundamental tension in the miner-as-Bitcoin-treasury model: these companies are simultaneously producers and holders, creating natural selling pressure that pure treasury companies like Strategy or Metaplanet do not face. Metaplanet's ability to generate yield on its holdings through options strategies while continuing to accumulate represents a more sophisticated and arguably more sustainable model — though it carries its own risks, particularly if Bitcoin experiences a deeper or longer correction than anticipated.

The broader market implication of this divergence is nuanced. The selling pressure from miners and distressed treasury companies is real and measurable — collectively, these entities have moved well over 15,000 BTC to exchanges in recent months ^[1] — but it is being partially absorbed by conviction accumulators on the other side of the order book. The net effect on price depends on the relative velocity of those flows. More importantly, the structural reduction in miner BTC reserves removes future overhang from the market: coins sold to service debt today cannot be sold again tomorrow. Paradoxically, forced liquidations, painful as they are in the short term, can contribute to a cleaner, less leveraged holder base over time.

Key Takeaways

• Miners under margin pressure are not ideologically abandoning Bitcoin — MARA's $1.1 billion sale and Riot's ongoing disposals are debt-management exercises, not conviction shifts, but they create measurable near-term selling pressure that investors should factor into price expectations ^[1][2].
• Metaplanet's rise to the #3 corporate holder position signals that conviction accumulation continues even at a loss — the firm's 32% unrealized drawdown has not altered its strategy, distinguishing it clearly from fair-weather treasury companies ^[3].
• The corporate Bitcoin treasury model is bifurcating into two archetypes: operationally-funded accumulators with genuine long-term conviction, and leveraged opportunists who built positions during rising markets but lack the balance sheet resilience to hold through a correction.
• Over 1.16 million BTC — more than 5% of total supply — remains locked in public company treasuries ^[2], meaning institutional Bitcoin adoption remains structurally significant even as individual participants rotate in and out.
• Delisting risks at miners like Cango and Canaan highlight an underappreciated systemic risk: if mining-linked Bitcoin treasury companies face capital markets pressure, forced selling could accelerate in ways that are difficult to predict from on-chain data alone ^[1].