From the ashes of 2017 to the fluidity of DeFi, I never thought I'd see a mining company promise returns that weren't really there. But here we are, staring at a Bernstein report that did what few analysts dare: it called out the elephant in the colocation room.
In the quiet hours of a recent research note, a single line cracked the facade of Bitcoin miners’ great AI pivot. Core Scientific, the poster child for mining-to-AI colocation, was proudly touting its partnership with CoreWeave as a revenue lifeline. But Bernstein saw what the market missed: the returns were twisted, not earned. The numbers looked healthy only because CoreWeave’s own financing structure inflated the payout. Strip out the capital injection, and the underlying hosting margins looked more like a slow bleed than a golden goose.
This isn’t just about one company. It’s a window into how narrative architecture can mask financial fragility—and how even the most bullish stories can collapse when the scaffolding is revealed.
Context: The Rise and Pivot of the Mining Behemoth
Core Scientific emerged from the 2021 mining boom as one of the largest publicly traded Bitcoin miners in North America, with a fleet of ASICs chewing through energy in Texas and beyond. Then the 2022 crash hit. Energy costs soared, Bitcoin prices cratered, and many miners went bankrupt or sold their rigs. Core Scientific filed for Chapter 11 in December 2022, burdened by debt and overleveraged expansion.
But the company didn’t die. It restructured, and in a stroke of narrative genius, pivoted to AI colocation. Instead of turning off its data centers, it repurposed them for high-performance computing. The market loved it. The stock surged. The narrative was that miners were not just commodity plays; they were potential AI infrastructure providers, riding the wave of generative AI demand. Core Scientific secured a $1.2 billion, 12-year contract with CoreWeave, a cloud AI startup backed by heavyweights like Nvidia.
But here’s the twist Bernstein uncovered: that contract’s returns were distorted by CoreWeave’s financing. CoreWeave wasn’t just paying for hosting; it was leveraging its own massive debt and equity raises to offer premium prices that didn’t reflect market rates. The “profit” Core Scientific reported was, in part, a phantom—a financial mirage born from the AI frenzy, not from sustainable operations.
Core: The Narrative Mechanism of Distorted Returns
From a sociological lens, this is a classic case of “narrative inflation.” Markets don’t price assets—they price stories. And the story of miners turning into AI landlords is one of the most compelling in the current bear market. It offers hope, diversification, and a bridge to the next bull run. But stories need backing data, and that data is often incomplete or self-referential.
Bernstein’s analysis essentially applied a forensic accounting framework to the narrative. They examined the contract’s cash flows, the counterparty risk of CoreWeave (which is not a cash-flow positive company, but a venture-backed startup burning capital), and the hidden clause that made Core Scientific’s returns look 2x better than they were in a market-neutral scenario.
I’ve audited similar structures before. In 2020, during the DeFi summer, I saw yield farming strategies that boasted 1000% APRs. But when you stripped out the token inflation and the recursive lending loops, the real yield was often negative. The same mechanism is at play here: an artificially supported return that creates a feedback loop of enthusiasm, but eventually reverts to mean.
The market sentiment around Core Scientific had been overwhelmingly bullish. Stock price up 150% year-to-date. Social media buzz about “miners as AI plays.” But sentiment without structural verification is just noise. Bernstein’s report was a splash of cold water, and it’s already starting to ripple.
Contrarian Angle: The Distortion Might Be the Only Reason It Works
Now for the contrarian view—the one that makes me pause before joining the bearish chorus. What if the distortion is not a bug, but a feature? CoreWeave’s financing isn’t a Ponzi scheme; it’s a purposeful subsidy from AI capital to secure compute capacity in a scarce market. The premium pays for reliability, speed, and the guarantee that Core Scientific will keep its data centers running for AI workloads instead of idling them.
In other words, the “distortion” is actually a premium for optionality. CoreWeave is paying extra to lock in compute for years, knowing that the cost of building their own data centers would be even higher. Core Scientific, in turn, gets a cash injection that could fund debt repayment and ASIC upgrades.
The real risk is not that the returns are distorted, but that they are concentrated and non-transferable. If CoreWeave’s own funding dries up, the entire contract collapses. And the market hasn’t priced that tail risk. The blind spot here is that investors cheered the revenue without questioning the source. They saw the number and assumed it was organic, when in fact it was chemically engineered.
Takeaway: Next Narrative and Survival Signals
In a bear market, survival matters more than gains. The question every mining stock holder must ask now is: are my assets safe? For Core Scientific, the answer may still be yes—but only if they diversify their client base and reveal the real (unwarped) unit economics. If they don’t, the narrative decay will accelerate, and we’ll see another chapter in the history of miners who soared on AI whispers and crashed on audit truths.
From the ashes of 2017 to the fluidity of DeFi, and now the ghost of the AI pivot, the lesson repeats: shake the narrative tree, and see which returns fall—and which were never real.
The next narrative? I’m watching for a new class of “transparent colocation” contracts, where hosting fees are pegged to cost-plus, not venture subsidies. That’s the only architecture that survives the scrutiny of a bear market. Until then, trust the data, not the story.