Skip to content
Contact
Events | Research

Bitcoin Mining Financing: Why Profitable Miners Still Run Out of Cash

By Andrea Maria Cosentino, Advisor

Bitcoin Mining Financing: Why Profitable Miners Still Run Out of Cash

I flew to Miami in July expecting a crypto conference. Mining Disrupt turned out to be something closer to a machinery fair. Transformers, immersion tanks, container shells, customs brokers, firmware vendors, and a surprising number of people who can fix a hash board in under an hour. On stage, the conversation ran to megawatts, curtailment revenue and whether a given site can honestly market itself to AI customers. The Bitcoin price barely came up. After three days of walking the halls, one thought kept returning. Plenty of the operators I met run tight, well-managed, profitable sites, and still spend half the year short of cash. The problem is that Bitcoin mining financing was designed for a different kind of company, one whose revenue arrives when the invoices do.

Bitcoin mining

 

Bitcoin mining profitability is under pressure

Start with hashprice, the industry’s shorthand for revenue per petahash per second per day. Through July 2026 it hovered between $29 and $32, a level last seen in the depressed months after COVID, except that today the network runs close to a full zetahash of far more efficient machines. CoinShares’ Q1 2026 mining report put the weighted average cash cost of producing a single bitcoin among listed miners at about $79,995, and reckoned that roughly a fifth of the global fleet loses money at current revenue levels. Around the turn of the year the network logged three consecutive negative difficulty adjustments, a streak last seen in July 2022, which is what capitulation looks like in protocol data.

Under that kind of pressure, miners sold. Listed operators moved more than 32,000 BTC off their balance sheets in the first quarter of 2026, a bigger quarterly liquidation than anything recorded during the Terra collapse, and more than they sold across the whole of 2025. Core Scientific let go of around 1,900 BTC in January. Bitdeer emptied its treasury in February. MARA, which had committed in July 2024 to holding everything it mined, changed course and sold 15,133 BTC for roughly $1.1 billion in March.

One detail from the MARA episode stayed with me. Part of the pressure came from a $350 million bitcoin-backed credit line whose loan-to-value ratio crept toward 87% as the price slid toward $68,000. That facility existed to help the company keep its bitcoin through a rough patch. When the rough patch arrived, it forced the selling instead.

 

Why Bitcoin miners have a cash flow problem

The tempting conclusion from all of this is that mining is dying. Spend three days around the people who actually run these sites and that reading gets hard to sustain. What I saw looks more like a cash flow mismatch playing out at industrial scale.

Mining income drips in continuously, block by block. Costs land in lumps. The power bill comes on the first of the month, the hosting deposit is due before the racks are energised, the repair invoice shows up whenever a board decides to die. A site can finish the year comfortably in profit and still spend one bad week selling coins at the bottom because a payment cannot wait. Multiply that across thousands of operators and you get a quarter that looks like capitulation, produced in part by businesses that simply ran out of runway between one difficulty adjustment and the next.

Capital does exist in this industry. It just pools at the top. Galaxy writes bitcoin-collateralised facilities with minimums around $1 million. The AI money, over $70 billion in announced contracts across the public mining sector by CoinShares’ count, flows to operators with the land, fibre and cooling that hyperscalers demand, and the equity market has already priced the split: miners with secured HPC contracts trade around 12.3 times forward sales, while the pure-play names sit near 5.9 times.

Most of the people I met in Miami live under neither umbrella. They run ten machines, or fifty, or five hundred. Institutional credit desks rarely look at tickets that small. Consumer lenders have no way to read a mining operation. DeFi pools will typically only touch collateral they can liquidate quickly. So a whole tier of operationally credible businesses ends up financing itself largely by selling the coins it produces as it produces them.

What better Bitcoin miner financing could look like

The first is a facility that advances working capital against verified production. The standard bitcoin-backed loan asks the borrower to already be wealthy: post coins, receive dollars. The miner who most needs credit is still selling everything to keep the lights on. A production-based line could size itself on hashrate, uptime, machine efficiency, power cost and pool payout history, all of which are observable in close to real time, and then split each payout three ways: operating cash, loan service, and a reserve of bitcoin that stays with the miner. Give an operator twelve months on a structure like that and they finish the year holding coins they would otherwise have sold in week one.

The second is financing placed where the upgrade actually happens. The most consequential moment in a small miner’s commercial life is the jump from machines running above 25 joules per terahash to hardware below 15, which roughly doubles revenue per megawatt-hour at current hashprice. That transaction runs through equipment resellers and recyclers rather than banks. A trade-in structure, where the old fleet is valued as the deposit, the lender funds the balance and the new machines stand as collateral until repaid, turns a hardware sale into a credit product at the moment of highest intent. The reseller closes more deals. The miner upgrades a year or two earlier than retained earnings would allow.

The third is a facility designed around the drawdown rather than in spite of it. MARA at 87% loan-to-value should be required reading for anyone structuring miner credit. A lender courting this market should be able to answer one question in plain language: what happens to my collateral when the price falls 30%? Pay funds directly to the hosting provider or power supplier. Amortise from production instead of ballooning at maturity. Let the collateral cushion grow as the loan ages. Every operator in that expo hall has watched a leveraged neighbour get liquidated, and in that crowd a boring, transparent structure is a sales pitch in itself.

There is also a ladder hiding inside those three products. A first facility at deployment. Working capital once production stabilises. Trade-in credit at the first upgrade. A bitcoin reserve that compounds until the borrower qualifies for conventional institutional terms. Each rung generates the underwriting data for the next one, and the lender who starts at machine one ends up owning the dataset that every future competitor will wish it had.

 

The fourteen-day problem

CoinShares expects hashprice to recover toward $37 if bitcoin reclaims $100,000. It might. I would rather build for the market where it doesn’t, because the structural forces stay in place either way. The halving keeps compressing margins on a fixed schedule. The AI bifurcation keeps pulling institutional capital toward the fifty largest operators. And underneath them, thousands of real businesses keep producing revenue on a clock that no lender has bothered to study properly.

Difficulty resets roughly every fourteen days. Bills keep their own calendar. Somewhere in the gap between those two clocks sits a lending business nobody has properly built, and after a week in Miami I am fairly sure the miners would queue for it.

Sources