When Miners Abandon the Vault: How Economic Pressure Is Reshaping Bitcoin’s Treasury Landscape
The calculus just flipped. For years, Bitcoin miners treated newly mined coins like sacred reserves-a “never sell” ethos that anchored conviction and long-term belief. But 2025 threw that playbook in the trash. As mining economics compressed into their tightest margins on record, the industry pivoted from treasury accumulation to active treasury management, and the implications for Bitcoin’s structural dynamics are anything but subtle.
This isn’t just miners getting desperate. It’s the canary in the coal mine signaling that the Bitcoin network has entered a new competitive era-one where energy costs, infrastructure depreciation, and the relentless rise of network difficulty have made passive holding economically irrational for industrial operators. The shift from hodling to selling, from accumulation to optimization, tells us something crucial: the infrastructure backbone holding Bitcoin supply off the market for years is now becoming a source of flow. And when large structural holders pivot their behavior, everything downstream changes.
Key Takeaways
Subscribe to our Social Media for Exclusive Crypto News and Insights 24/7!
• Bitcoin miner profitability compressed to historic lows in 2025, with all-in production costs reaching $137,800 per BTC amid 1 ZH/s hashrate, pressuring treasury policies and forcing strategic asset liquidation.
• Marathon Holdings (MARA) expanded bitcoin sales authorization in 2026, signaling institutional miners’ shift from passive treasury accumulation to active balance-sheet optimization amid tighter margins.
• Network hashrate surge and halving impact created structural supply overhang, as April 2024’s block reward reduction and rising difficulty reduced miner revenue per unit of computational power by half.
• Industry consolidation accelerated post-halving, concentrating hashpower-Foundry USA and MARA Pool now control 38% of global hashpower, amplifying individual policy decisions’ market impact.
• Miner diversification into AI/HPC workloads signals infrastructure reallocation away from pure Bitcoin mining, with CoreWeave’s $9 billion acquisition of Core Scientific marking the sector’s pivot toward heterogeneous compute infrastructure.
The Economics Don’t Lie: Margin Compression at the Mining Layer
Here’s the brutal math that nobody wanted to see coming. According to CryptoRank’s December 2025 analysis, the average cash cost for publicly listed mining companies to produce one Bitcoin had climbed to approximately $74,600[7]. Sounds manageable until you factor in depreciation and stock-based compensation-then the all-in cost balloons to roughly $137,800 per coin[7]. When you’re mining Bitcoin at that production cost and watching prices hover (and historically average around much higher levels), you’re not making a killing. You’re playing a cash-flow management game just to stay solvent.
This didn’t happen overnight. The April 2024 halving cut block rewards from 6.25 BTC to 3.125 BTC-a 50% overnight reduction in mining revenue per block[6]. The network didn’t panic or collapse; instead, hashrate continued its relentless ascent. By December 2025, Bitcoin’s total network hashrate officially surpassed 1 ZH/s (zettahash) for the first time[7]. More computational power competing for the same halved rewards means individual miners’ share of the pie shrinks proportionally. It’s basic economics: when supply of an input (block rewards) drops while demand (hashrate) skyrockets, unit economics deteriorate fast.
The revenue picture tells the story. Monthly aggregate miner revenue hit $2 billion in March 2024 before the halving shock[5]. Post-halving, revenues compressed into a range between roughly $1.19 billion (April 2024) and $1.63 billion (August 2025), with the December 2025 figure settling well below 2024 comparables[6]. That’s a structural headwind, not a temporary dip. And when margins compress this severely, balance-sheet strategies change.
Marathon Holdings’ decision in their 2026 annual filing is the canary-and it’s definitely not singing a happy tune[1]. The company disclosed that it would expand its crypto management strategy to permit sales of bitcoin held on its balance sheet, a marked departure from their previous policy of selling only newly mined production[1]. As of December 31, 2025, Marathon held 53,822 Bitcoin valued at approximately $4.7 billion (based on year-end spot price of $87,498)[1]. That’s serious dry powder.
What changed their mind? Marathon mined 8,799 Bitcoin in 2025, down from 9,430 in 2024-a direct reflection of both the halving and rising network difficulty[1]. The company explicitly noted that “revenue remains tied to bitcoin’s market price, while costs such as electricity, infrastructure, and financing remain fixed or rising”[1]. Translation: when you’re generating less Bitcoin from the same operational footprint and your costs aren’t shrinking, you eventually have to monetize reserves to fund operations or return capital to shareholders. That’s not capitulation-that’s arithmetic.
Treasury Policy Inversion: From Accumulation to Optimization
The philosophical shift here matters more than the immediate flow impact. For years, the mining industry’s self-narrative centered on long-term conviction. Miners were the original HODLers-believers so committed to Bitcoin’s future that they’d hold produced coins indefinitely, creating a structural bid under the supply curve. It was a powerful market microstructure dynamic: a whole industry class essentially removing supply from circulation.
Now that’s reversing. BitFuFu’s (NASDAQ: FUFU) operational decisions in late 2025 illustrate the broader repositioning[3]. In November 2025, the company “opportunistically” sold 205 Bitcoin, reducing its treasury to 1,764 BTC by month end[3]. The language is clinical-”opportunistically”-but the substance is unmistakable: miners are now watching price levels and making tactical liquidation decisions rather than running a passive accumulation strategy[3].
This transition has profound implications. When miners were automatic buyers (via hodling), their operating revenues created a persistent downward pressure on spot prices-they had to sell just enough to cover operational costs, leaving the vast majority of production in cold storage. That structural dynamic supported a narrative of long-term scarcity and conviction-driven accumulation. Now, with tighter margins, the calculus shifts. Miners become selective sellers, tactical traders, and balance-sheet optimizers rather than passive holders.
Marathon’s strategic shift also reveals something about institutional maturity. The company’s 2025 policy allowed sales only from newly mined production[1]. The 2026 expansion to permit sales from existing treasury holdings[1] suggests management has concluded that passive treasury accumulation is no longer defensible to shareholders when margins compress this significantly. This is what capital discipline looks like: acknowledging that opportunity cost of holding unproductive assets (in terms of yield or operational returns) exceeds the optionality value of maintaining a full treasury.
The tension is real though. Marathon acknowledged recording a $422.2 million decrease in fair value of its Bitcoin holdings during 2025 as prices fluctuated[1]. Holding large BTC treasuries amplifies gains in bull markets but magnifies balance-sheet pressure during downturns[1]. In other words: treasuries create volatility in GAAP earnings, something institutional shareholders increasingly question. A treasury policy shift that allows monetization smooths earnings volatility and gives management more tactical flexibility-exactly what institutional boards demand.
Industry Consolidation: When Winners Centralize, Market Structure Changes
The post-halving landscape didn’t produce a level playing field. It produced exactly the opposite. According to AlixPartners’ comprehensive post-halving analysis, industry consolidation accelerated significantly, with smaller miners exiting the market due to tighter margins while larger firms capitalized on mergers and acquisitions to scale operations and secure power access[4].
The hashpower concentration is stark. The top pools-Foundry USA and MARA Pool-now account for over 38% of global Bitcoin hashpower[4]. That’s not trivial. When more than a third of the network’s computational resources are concentrated among two entities, individual policy decisions carry outsized systemic weight. When Marathon or Foundry-backed operations shift their treasury or liquidation policies, it moves the needle on actual on-chain flow dynamics.
This centralization story has historical parallels. In traditional commodities, when production capacity consolidates into fewer hands, those producers gain pricing power and operational flexibility that smaller competitors can’t match. Bitcoin mining is following a similar arc. Larger miners can:
- Negotiate bulk power deals at lower marginal costs due to scale
- Deploy next-generation ASICs faster due to capital availability
- Diversify into adjacent revenue streams (AI compute, hosting services) to offset mining margin compression
- Maintain balance sheets that absorb commodity price volatility without existential pressure
Smaller miners face a different reality. They lack the capital to upgrade ASIC fleets quickly, can’t negotiate favorable power contracts, and can’t afford to hold large treasuries through bear markets. Result: they exit, get acquired, or get forced into tactical liquidation mode just to survive. The ones that exit create selling pressure exactly when they’re most distressed-typically near cycle lows.
But here’s where it gets interesting structurally. CoreWeave’s July 2025 all-stock acquisition of Bitcoin miner Core Scientific for approximately $9 billion[7] signals something broader: the value proposition of mining infrastructure itself is shifting away from pure Bitcoin production. Core Scientific brought 1.3 gigawatts of scarce power capacity to the table[7]. That’s not being repurposed primarily for Bitcoin mining anymore-it’s being redeployed toward AI and high-performance computing workloads, where margins and utilization rates are currently superior.
This is the real strategic pivot. Miners aren’t just selling treasury holdings to manage balance sheets. They’re fundamentally reconceptualizing what their infrastructure is for. Bitcoin mining is becoming one application among many, not the primary use case. That’s a structural shift in how mining operators think about capital allocation and return optimization. When infrastructure owners treat Bitcoin mining as one revenue stream among many rather than the core business, their behavior around Bitcoin holdings changes materially.
The All-In Cost Reckoning: $137,800 Per Coin Rewrites the Playbook
Let’s drill into that all-in cost figure, because it’s where the rubber meets the road for miner behavior.
According to CryptoRank’s December 2025 data, while average cash costs for publicly listed miners hit $74,600 per BTC, the all-in cost including depreciation and stock-based compensation reached approximately $137,800 per coin[7]. That second figure is the one that matters for long-term treasury and hedging decisions.
Why? Because all-in cost includes the non-cash but very real economic cost of equipment depreciation. When you’ve invested capital in ASIC hardware with a 3-5 year useful life, that depreciation expense is real to shareholders even if it doesn’t hit the cash P&L immediately. Similarly, stock-based compensation to attract and retain technical talent represents real economic dilution. Including both in production cost gives you a better view of the true economic burden of mining operations.
So here’s the practical implication: if your all-in cost is $137,800 and you’re holding Bitcoin, you’re implicitly accepting that the price needs to move significantly higher than that level just to break even on a total economic basis. That’s a psychologically different position than when all-in costs were $30,000-50,000 (as they were in earlier cycles). When the barrier to profitability is that high, miners become less ideological about holding and more pragmatic about liquidity management.
Marathon held 53,822 BTC valued at roughly $4.7 billion as of December 31, 2025[1]. That treasury represents roughly 6 years of their current production run rate (8,799 BTC in 2025[1]). From a corporate finance perspective, holding 6 years of production as inventory on the balance sheet while facing compressed margins is aggressive. Most industrial producers target 3-6 months of inventory. Bitcoin miners historically held much higher multiples because they believed in long-term appreciation. But when you’re compressed on margins and facing shareholder scrutiny, that conviction gets tested.
The Network Difficulty Ratchet: Why It Only Goes Up
The difficulty adjustment mechanism is one of Bitcoin’s most elegant design features-and one of the most brutal realities for miners. Every 2,016 blocks (roughly 2 weeks), the network recalibrates mining difficulty to maintain a consistent ~10-minute block time[5]. When hashrate increases, difficulty increases proportionally. When hashrate decreases, difficulty decreases. In theory, it’s a beautiful self-balancing system.
In practice, it creates a one-way ratchet that only goes up during growth periods. From April 2024 through December 2025, difficulty reached new all-time highs repeatedly, forcing older ASIC hardware into obsolescence and making older equipment uneconomical to operate even during high-price environments. That’s why you saw a resurgence in older models like Whatsminer M32 units (launched August 2020, ~50 J/TH efficiency) and Antminer S9s (released 2017, ~93 J/TH efficiency)-they were being dragged out of retirement just to add hashrate during the bitcoin price rally that followed Trump’s November 2024 election[6].
This creates a perverse dynamic: older equipment gets redeployed exactly when margins are worst, because it’s so inefficient that it barely breaks even. New equipment (like the latest Antminer S21 and T21 devices, or Whatsminer M76 and M78 immersion-cooled units) commands premium pricing and represents new capital expenditure precisely when capital is tight. The result is that marginal mining remains economically irrational for extended periods, but the infrastructure persists because fully depreciated older hardware still covers marginal operating costs.
Marathon’s production decline from 9,430 BTC (2024) to 8,799 BTC (2025) despite running operations all year tells you something about that dynamic[1]. They didn’t shut down; their hash contribution simply declined as a percentage of a larger network. That’s the squeeze: when the network grows faster than your capital deployment can match, your market share shrinks even while you’re investing.
Miner Selling as a Leading Indicator: The Supply Cycle Shifts
Here’s what most market participants miss: miner selling behavior is a leading indicator for supply cycle dynamics, not a lagging one. When structural holders change their liquidation behavior, it often precedes broader market shifts.
Think about what Marathon’s policy change signals. A major, publicly listed, institutional-grade miner is saying: “We no longer believe passive treasury accumulation makes economic sense.” If they’re making that call, you can bet other miners are making similar internal decisions. BitFuFu’s “opportunistic” November sales[3] suggest others are already acting on it.
This creates a two-layer supply dynamic:
- Baseline production flow: Roughly 450 BTC per day post-halving[6] hitting the market from active mining and immediate operational liquidation
- Discretionary treasury liquidation: Increasingly tactical selling from miners managing balance sheets and optimizing returns
When layer two activates-when miners shift from passive holders to active liquidators-it changes the supply absorption equation. It’s not a massive daily addition (baseline production is what it is), but it’s a shift in the composition of supply and the predictability of when that supply hits the market.
Historical precedent matters here. After the 2016 halving, we saw similar compression in mining margins, and several major mining operations shifted their strategies. The result was modestly increased selling pressure through 2017 before a surge in Bitcoin prices eventually compressed mining difficulty again. We’re not necessarily repeating that cycle, but we’re not on a fundamentally different one either.
The AI Pivot: Mining Infrastructure’s Identity Crisis
The CoreWeave acquisition of Core Scientific for $9 billion deserves serious attention because it represents something structural[7]. This wasn’t a mining company acquiring another mining company. This was an AI infrastructure provider using vast capital to acquire the power infrastructure that Bitcoin miners had built.
Core Scientific brought 1.3 gigawatts of power capacity to CoreWeave[7]. That’s a substantial amount of physical infrastructure-data center space, electrical connectivity, cooling systems, all optimized for high-density compute. In the current market environment, AI and high-performance computing workloads generate better margins than Bitcoin mining. So CoreWeave’s decision to acquire Core Scientific was fundamentally about redirecting that power capacity toward higher-margin applications.
This signals something important: Bitcoin mining as an application is losing the infrastructure arms race. The hardware, power contracts, and operational expertise that miners spent years developing are being repurposed because the ROI on AI compute is currently superior. That’s not unique to Bitcoin-it’s happening across commoditized hardware-based businesses when superior applications emerge.
For miners still operating pure Bitcoin operations, that means they’re competing against infrastructure operators who can flexibly allocate power and hardware between Bitcoin mining and AI/HPC workloads, choosing whichever offers superior returns on any given day. That’s a competitive pressure that didn’t exist two years ago. It’s part of why margin compression is hitting so hard: miners aren’t just competing against other Bitcoin miners; they’re competing against operators who can instantly reallocate to completely different workloads.
Marathon’s strategic shift to allow treasury sales takes on additional context here. If infrastructure is being repurposed and redirected, holding Bitcoin on the balance sheet becomes increasingly costly from an opportunity-cost perspective. You’re not just forgoing Bitcoin price appreciation (which you’d capture anyway if you held production); you’re forgoing the ability to redeploy capital into higher-margin applications.
Market Timing and Supply Absorption: The Missing Bid
Here’s where the positioning gets interesting. Marathon held 53,822 BTC valued at roughly $4.7 billion as of year-end 2025[1]. That’s material supply. But it’s not being dumped into spot markets-it’s being managed strategically, which is exactly what you’d expect from an institutional operator.
The question worth watching: how does the market absorb this supply if miners shift from passive holders to selective liquidators?
Historically, the absorption mechanism worked like this:
- Miners produce 450 BTC daily (post-halving)
- Miners need cash for operations, so they sell enough production to cover costs
- Miners hold excess production
- Over time, production accumulates, creating a gradually shrinking available supply from miners
Now the mechanism shifts:
- Miners produce 450 BTC daily
- Miners need cash, so they sell production
- Miners also strategically liquidate treasury holdings
- The net effect is larger, more consistent supply hitting the market
This isn’t necessarily bearish-it depends entirely on demand absorption. If there’s strong institutional demand (via Bitcoin ETFs, corporate treasuries, etc.), that supply gets absorbed easily. If demand is tepid or event-driven, it creates periods of price pressure.
What we don’t see in the available data is a dramatic flood of miner selling that’s crashed the market. Marathon’s strategy expansion was announced in a regulatory filing; if they were dumping $4.7 billion of Bitcoin, we’d see that in on-chain metrics. So miners seem to be managing this strategically rather than panic-selling.
That discipline actually matters for market structure. It suggests that even compressed miners aren’t capitulating-they’re optimizing. And optimization is a much more manageable supply dynamic than capitulation.
The Consolidation Implication: Fewer, Larger, More Strategic
AlixPartners’ observation that larger firms capitalized on M&A to scale operations and secure power access[4] deserves emphasis because it changes the microstructure of how mining decisions get made.
When mining was more fragmented, individual operators made decisions based on immediate cash needs and personal conviction. Now, with consolidation around Foundry USA and MARA Pool controlling 38% of global hashpower[4], individual decisions get made by institutional boards with fiduciary duties to stakeholders. That means:
- Decisions are more deliberate and less emotional
- Balance sheets matter more than individual operator conviction
- Coordinated policy changes can occur (not in a collusive sense, but in a correlated sense)
- Institutional investors own stakes and influence treasury decisions
This is actually stabilizing in some ways (less panic selling, more rational long-term planning) and destabilizing in others (when a few large players change direction simultaneously, it matters more to the market). The net effect is that mining behavior becomes more institutional and less idiosyncratic.
Historical Context: When Miners Have Shifted Strategy Before
The 2016 halving compressed mining margins severely. Antpool’s Chen Haocheng and other pool operators made similar strategic decisions to sell more aggressively rather than accumulate[5]. The result was measurable increase in supply flow through 2016-2017, but it didn’t prevent a bull run-because demand dynamics eventually overwhelmed supply considerations.
The 2020 halving was less dramatic on the miner selling front because the COVID rally in 2020 was so strong that it offset margin compression quickly. By late 2020, mining was profitable again on a fully-burdened basis and miners went back into accumulation mode.
This cycle (2024-2025) is different because the margin compression is hitting harder and lasting longer than previous post-halving periods. That’s why we’re seeing more dramatic policy shifts from major miners. It’s not panic-it’s adaptation to genuinely tighter economics.
Looking Forward: Structural Supply Absorption Questions
The next critical question is how Bitcoin’s demand ecosystem absorbs this shifted supply dynamic.
Bitcoin ETF inflows have been meaningful in 2025[2], and institutional adoption continues growing. If that demand remains strong, miner liquidation becomes almost immaterial-institutional buyers will absorb it happily. If institutional demand dries up or gets event-driven, miner selling becomes tactically important for price levels.
The other variable is Bitcoin’s price itself. If spot price moves materially higher, all-in costs look better and miners get less urgent about liquidation. If price languishes, margin pressure intensifies and tactical selling likely accelerates. The relationship is circular: price pressure begets more selling (from compressed miners), which creates supply headwinds, which potentially pressures price further.
That’s the structural imbalance worth watching: miners’ treasury policies are now increasingly price-responsive and margin-responsive rather than conviction-driven. That creates potential for feedback loops.
The Unspoken Message: Infrastructure Maturation and Role Consolidation
Marathon’s treasury policy shift and the broader miner pivot into AI/HPC signals something philosophical: Bitcoin mining is maturing from a speculative, conviction-driven activity into a professional infrastructure business. And infrastructure businesses optimize for cash flow and return on invested capital, not for accumulation and belief.
That’s not bearish. It’s just different. It means miners are becoming more like utilities and less like believers. That’s actually healthy for long-term market maturation-it removes the emotional volatility from a major supply component and replaces it with economic logic.
But it also means the structural narrative around “miners as long-term holders creating supply scarcity” is becoming less relevant. Miners are becoming supply managers, not supply hoarders.
The net result: Bitcoin’s supply dynamics are becoming more predictable and less structurally supportive to the “artificial scarcity” narrative. But they’re also becoming more mature and professional. That’s a trade-off worth understanding as you size your positions.
- https://bitcoinmagazine.com/news/these-two-bitcoin-sell-their-bitcoin
- https://www.intelmarketresearch.com/bitcoin-miner-market-11021
- https://coingeek.com/btc-miners-see-little-light-ahead-at-the-end-of-2025-tunnel/
- https://www.alixpartners.com/insights/102m1on/2025-bitcoin-miners-landscape/
- https://aminagroup.com/research/post-halving-bitcoin-miners-landscape/
- https://forklog.com/en/bitcoin-mining-in-2025-the-harshest-profitability-squeeze-on-record-amid-all-time-highs/
- https://wublock.substack.com/p/top-10-mining-news-of-2025-miners
The next move in miner supply dynamics won’t be determined by conviction-it’ll be determined by the margin spread between Bitcoin production costs and spot price, and increasingly by competition with AI infrastructure for the same power capacity.
