While the market chases the next Bitcoin price milestone, the instruments designed to capture its upside are quietly rewriting their own DNA. The 90-day correlation data released by Tom Lee’s team reveals a stark truth that most investors are not ready to accept: the stocks once considered the ultimate proxy for Bitcoin exposure are now moving to a different rhythm. Core Scientific, a name that once rode the BTC wave, now shows a mere 16% correlation with the underlying asset. Riot Platforms sits at 31%, IREN at 33%. Meanwhile, MicroStrategy—a company that does not mine at all—holds a commanding 78% correlation. This is not a statistical anomaly. It is a structural reclassification of an entire asset class.
Context: The Landscape of Crypto Equity Proxies
Tom Lee’s analysis covered 17 publicly traded stocks with market capitalizations exceeding $2 billion. The universe includes pure-play treasury companies like MicroStrategy, exchanges like Coinbase, and a growing cohort of mining firms that are rapidly transforming into AI infrastructure providers. The metric used is a 90-day rolling correlation with Bitcoin and Ethereum prices—a standard tool for gauging how closely these equities track the underlying digital assets. Historically, mining stocks were considered high-beta proxies: if Bitcoin moved 10%, a miner’s stock might move 20% due to operational leverage. But that relationship is dissolving.
Data from the first quarter of 2025 shows a clear divergence. MicroStrategy, with its massive BTC treasury, remains the gold standard for equity-based Bitcoin exposure. Coinbase, at 74% correlation with ETH, reflects its role as the primary on-ramp for retail and institutional crypto activity. But the mining cohort tells a different story. BitMine leads the ETH correlation list at 80%—yet Tom Lee is its chairman, introducing a conflict of interest that demands scrutiny. The real story lies in the miners that have embraced AI compute leasing.
Core: The Structural Shift from Hashrate to Compute
My experience auditing DeFi protocols during the 2020 summer taught me that when revenue streams shift, correlations follow. The same principle applies here. Mining companies are no longer pure plays on Bitcoin’s hashrate. They are becoming landlords of power and data center space, renting out their infrastructure to AI firms that need cheap, reliable compute. The transformation is not incremental—it is fundamental.
Core Scientific, after emerging from Chapter 11 bankruptcy, now derives a significant portion of its revenue from AI hosting contracts. TeraWulf’s CFO recently stated that the company’s revenue will increasingly be driven by recurring contractual income from AI clients rather than volatile Bitcoin mining rewards. IREN, once a pure Bitcoin miner, now allocates over 60% of its computational capacity to AI workloads. These are not side projects; they are core business pivots.

The financial implications are clear. AI compute contracts offer stable, predictable cash flows compared to the brutal cyclicality of Bitcoin mining. But they also change the stock’s sensitivity. A miner’s share price now responds to AI demand forecasts, electricity costs, and data center utilization rates, not just the next Bitcoin halving or price rally. The 90-day correlation coefficients reflect this: as AI revenue share increases, BTC correlation drops. The relationship is inverse and monotonic.

Consider the numbers: MARA and CleanSpark, two miners that aggressively pivoted to AI, have collectively reported $851 million in losses from the transition. The market is pricing in the promise of future AI revenue, but the execution risk is real. Yet the narrative is compelling enough that investors continue to treat these stocks as crypto proxies, creating a dangerous mismatch between expectation and reality.
Contrarian Angle: The Decoupling Is Not a Market Mispricing
The conventional wisdom is that this decoupling is temporary—that once Bitcoin rallies, mining stocks will revert to their old beta. I believe the opposite is true. The decoupling is not a market inefficiency to be arbitraged away; it is a structural reclassification of an asset class. Mining firms are becoming AI infrastructure providers, and their equity should be valued as such. The market is slow to reprice because the old narrative—miners as Bitcoin proxies—is deeply embedded. But the data is unambiguous.
From speculative frenzy to institutional ledger. The same shift we saw in DeFi, where yield farming gave way to sustainable lending protocols, is now happening in the mining sector. The state does not compete; it absorbs. In this case, the AI industry is absorbing the mining infrastructure. The miners that survive will be those that lock in long-term AI contracts and manage their balance sheets conservatively. The ones that cling to pure Bitcoin mining will face margin compression as the network difficulty rises and the block reward halves.

There is a contrarian opportunity here: investors who recognize that mining stocks are now AI infrastructure plays can position themselves for a new cycle. But they must be clear-eyed about the risks. The AI narrative is hot, but it is also fragile. If AI demand cools, these miners could lose both their AI premium and their Bitcoin beta, leaving them in a valuation vacuum. The volatility is merely the tax on uncertainty.
Takeaway: Recalibrate Your Crypto Exposure Strategy
If your goal is to gain exposure to Bitcoin, the most efficient equity vehicle remains MicroStrategy. Its 78% correlation, combined with its leverage and treasury management, provides a direct, albeit risky, link to BTC price action. For Ethereum exposure, Coinbase offers a cleaner proxy, though regulatory risks and trading volume fluctuations add noise. Mining stocks, on the other hand, should no longer be considered crypto proxies. They are hybrid assets with a growing AI infrastructure component, and their valuation will increasingly depend on factors outside the crypto ecosystem.
Yields dissolve; infrastructure remains. The mining industry is undergoing a fundamental transformation, and the stocks that survive will be those that anchor themselves to the new AI economy. For the macro watcher, this is a classic case of structural change outpacing market perception. The smart money is already reclassifying. The question is whether the rest of the market will catch up before the next cycle begins.
Based on my research into CBDC architecture and monetary policy transmission, I have learned that when the underlying asset changes, the proxies must be re-evaluated. The same logic applies here. The Bitcoin mining stock of 2021 is not the same asset as the Bitcoin mining stock of 2025. The code enforces what contracts cannot—and in this case, the code of the market is rewriting the contract between miners and investors. Those who fail to see the shift will find themselves holding exposure to a narrative that no longer exists.