MARA did not sell Bitcoin because it stopped believing in Bitcoin. It sold Bitcoin because a 505-megawatt power plant in Ohio generates more revenue pointed at AI hyperscalers than at SHA-256 hashing. That is not a philosophical statement. That is an electricity bill.
Every outlet covered this as a crisis of conviction. They got the story wrong.
What the math actually shows
JPMorgan puts the average industry production cost at $77,000 per Bitcoin in Q1 2026. Hash price collapsed 66% from Bitcoin’s October peak. The 2024 halving cut block rewards in half. Network difficulty hit all-time highs. Efficient miners with sub-$0.05 power contracts can still turn a profit. Everyone else is somewhere between breakeven and underwater.
MARA is not the best-positioned operator. It reported a $1.26 billion net loss in Q1. It was carrying $3.3 billion in convertible debt. The Bitcoin it was mining was not covering the cost of the business built around mining it.
AI hosting currently generates more revenue per megawatt than Bitcoin mining at current hashprice levels. That sentence is not an opinion. It is the operational conclusion every major miner is arriving at independently. MARA, IREN, Core Scientific, Riot. The entire sector is asking the same question: what pays more per megawatt right now? The answer in 2026 is AI.
What MARA actually did
MARA sold 20,880 Bitcoin for $1.5 billion. It used $1 billion to retire 30% of its convertible debt. It deployed the rest into the Long Ridge Energy campus in Ohio. 505 megawatts. 1,600 acres. Plans to repurpose up to 90% of mining capacity for AI data center workloads.
CFO Salman Khan called Bitcoin “a source of strategic financial flexibility.” That is the key line. MARA is not abandoning Bitcoin as a store of value. It is treating Bitcoin as working capital. The same way a company sells inventory to fund a factory expansion, MARA sold Bitcoin to fund a power infrastructure pivot.
The comparison that actually matters
Cryptopolitan is using Strategy as the philosophical counterweight. Strategy bought 145,834 Bitcoin year-to-date. Saylor has not sold a coin. MARA believes, Strategy doubts, and so on.
That framing is wrong in a specific way worth naming.
Strategy is not a mining company. It has no electricity bill. No ASIC fleet. No hashrate exposure. It acquires Bitcoin using equity and debt raises priced against its own stock premium. When Strategy buys Bitcoin it is expressing a view on monetary debasement. When MARA sells Bitcoin it is responding to electricity costs, network difficulty, and debt obligations.
These are different businesses. Their Bitcoin decisions cannot be read as opposite sides of the same ideological argument.
What this tells you about mining in 2026
The operators who will still be mining Bitcoin in 2028 are the ones with sub-$0.05 power contracts and next-generation ASICs. Everyone else is either pivoting to AI, selling to better-positioned operators, or winding down.
MARA’s Long Ridge acquisition is a recognition that 505 megawatts is more valuable as AI compute than as mining capacity at current economics. The hyperscaler contracts it is signing pay in dollars regardless of what Bitcoin does.
That certainty is worth something when your alternative depends on an asset price staying above your production cost every single day.
Bitcoin did not fail MARA. Mining economics did.
