The EV/EBITDA multiple for Bitcoin miners with AI contracts is 12.3x. For pure miners, it is 5.9x. That is not a spread; it is a structural chasm. Over the past twelve months, hash price has dropped 50% from $53 to $31.8 per PH/s. Network hash rate has fallen 21% from 1.14 ZH/s to 900 EH/s. Yet WULF, IREN, and CIFR each rose over 100%. MARA fell 40%.
This is not a story about leverage. This is a story about asset repurposing.
Context: The Miner's Dilemma
Bitcoin mining is a commodity business. The moat is cheap power and scale. In 2025, that moat began to erode. The halving compressed margins. Hash price collapsed. Miners with marginal power costs faced an existential choice: shut down or pivot.
Enter the AI data center boom. Hyperscalers like Anthropic, OpenAI, and Google are desperate for power. They need 100 MW sites built yesterday. Bitcoin miners sit on exactly that: substations, fiber, cooling, and long-term power contracts. The pivot is not a technology innovation; it is a business model extension. Miners are transforming from "electricity-to-bitcoin" converters into "electricity-to-compute" brokers.
CoinShares estimates miners have signed over $70 billion in AI/HPC contracts. Riot's 20-year, $9.1 billion deal with Anthropic is the poster child. The market rewarded these contracts with a 2.1x EV multiple premium. Follow the gas. Always.
Core: The On-Chain Evidence Chain
Let me be precise. The data tells a clear hierarchy of value.
First, the hash price decline is not a blip. At $31.8/PH/s, the average miner with an all-in cost of $40/PH/s is losing money. Network hash rate dropping 21% confirms the marginal producers are shutting down. This is a forced deleveraging.
Second, the survivors are those with the lowest power tariffs. These are the same assets that attract AI clients. A 100 MW site with a 5-year fixed power purchase agreement at $0.03/kWh is worth more as a GPU cluster than as an ASIC farm. Code is law; math is evidence. The math favors the transition.
Third, the divergence in valuation multiples is not speculative. It is a rational discounting of future cash flows. A pure miner's revenue is a function of Bitcoin price and network difficulty—both volatile and unpredictable. An AI miner's revenue is a contracted stream of dollar-denominated payments over 10-20 years. The market is swapping volatility for visibility.
Contrarian: Correlation Is Not Causation
But the market is pricing in a perfect execution curve. It rarely exists.
Here is the blind spot: not all power contracts are created equal. Bitcoin miners typically use interruptible power—cheap but unreliable. AI data centers require 99.999% uptime. Converting a mine to a Tier III data center requires upgrades: liquid cooling, redundant fiber, and backup generators. That costs $10-20 million per 10 MW. The market may be underestimating the capex.
Second, the long-term contracts are not guaranteed. They are framework agreements with milestone payments. If the AI client's demand falters or a better GPU arrives, renegotiation risk is real. Core Scientific and CoreWeave had multiple contract adjustments. Volatility exposes leverage.
Third, the narrative is ahead of the revenue. WULF, IREN, and CIFR have seen their stocks double, but their GAAP earnings from AI are still a fraction of total revenue. The market is pricing a 2027 scenario in 2025. That creates a window for disappointment if Q3 earnings show slower-than-expected buildout.
Takeaway: The Next Quarter Signal
I will be watching two metrics in the coming weeks.
First, the ratio of contracted power capacity to actual deployed GPU capacity. A high ratio with low deployment indicates execution risk. Second, the proportion of AI revenue in each miner's income statement. If a miner claims an AI contract but shows no material revenue after two quarters, the market will reprice.
The hybrid miners—those that maintain a profitable Bitcoin mining operation while building AI capacity—are the safest bet. Pure miners without AI contracts are trading at a discount that may become a value trap if Bitcoin price does not rally to $126,000, the level needed to restore hash price to $59/PH/s.
Based on my forensic analysis of 50,000 wallets during the Terra collapse, I learned that narrative-driven markets collapse when the underlying data stops supporting the story. The same applies here. The pivot is real, but the valuation gap is pricing in a perfect landing. The next quarter will tell us whether the runway is long enough.
Follow the gas. Always.