Just looked at the latest mining data and something pretty significant is happening that most people are still sleeping on. Bitcoin miners aren't really bitcoin miners anymore - they're becoming AI companies, and it's happening faster than you'd think.

The numbers tell the story. Production costs hit nearly $80K per BTC in Q4 2025 while prices were hovering around $68-70K. That's roughly $19K in losses per coin, which obviously isn't sustainable. So what's the move? The entire sector is pivoting hard into AI and high-performance computing infrastructure. We're talking over $70 billion in cumulative contracts announced across public miners alone.

Core Scientific signed a $10.2 billion deal with CoreWeave. TeraWulf locked in $12.8 billion in HPC revenue. Hut 8 grabbed a $7 billion, 15-year lease for AI infrastructure. These aren't side projects anymore - by end of 2026, some of these companies could be pulling 70% of their revenue from AI, up from around 30% today. Core Scientific is already at 39%. That's a complete business model transformation.

Here's the economics part that actually makes sense. Bitcoin mining infrastructure costs roughly $700K-$1M per megawatt. AI infrastructure? $8-15 million per megawatt. Way higher capex, but the returns are completely different. AI contracts are promising margins above 85% with multi-year visibility. Bitcoin hash price hit around $28-30 per petahash per day in early March - miners need sub-$0.05/kWh electricity just to break even. The choice is pretty obvious.

How are they funding this? Two ways. First, debt - and we're talking infrastructure-scale debt loads now, not mining-scale. IREN carrying $3.7 billion in convertible notes. TeraWulf at $5.7 billion total debt. Cipher Digital's quarterly interest expense jumped from $3.2 million to $33.4 million in Q4 alone. These are bets that AI revenue materializes fast enough to service the obligations.

Second, bitcoin sales. Miners have collectively reduced their BTC treasuries by over 15,000 coins. Core Scientific dumped roughly 1,900 BTC worth $175 million in January and is planning to liquidate substantially all remaining holdings this quarter. Bitdeer went to zero in February. Riot sold 1,818 BTC worth $162 million in December. Even Marathon, the largest public holder with 53,822 BTC, quietly expanded its policy in March to authorize sales from its entire balance sheet. That's a signal.

Here's where it gets interesting though - these are the same companies securing the bitcoin network. When they stop mining because AI is more profitable, the network's security budget shrinks. Hashrate already peaked around 1,160 exahashes per second in October 2025 and has declined to roughly 920 EH/s with three consecutive negative difficulty adjustments. That's the first streak like that since July 2022.

The market gets it. Miners with secured HPC contracts trade at 12.3x next-twelve-month sales. Pure-play miners at 5.9x. Investors are paying more than double for the AI exposure, which just reinforces the incentive to pivot further.

CoinShares is forecasting hashrate reaching 1.8 zetahashes by end of 2026, but that depends on bitcoin recovering to $100K by year-end. Current price is around $74K, so we're not there yet. If prices stay below $80K, hash price continues falling and more miners exit. Below $70K triggers bigger capitulation.

Next-gen hardware like Bitmain's S23 and Bitdeer's SEALMINER A3 could roughly halve energy costs per bitcoin, but deploying them requires capital that most miners are directing toward AI instead.

The transformation is real. Bitcoin mining entered this cycle as companies that secured the network and accumulated bitcoin. It's exiting as companies that build AI data centers and sell bitcoin to fund them. Whether this is temporary or permanent depends on one variable: bitcoin's price. If it hits $100K, mining margins recover and the pivot slows. If it stays at current levels or lower, the transition accelerates and the mining sector as we knew it basically disappears into something else entirely.