The signal came not from a whitepaper or a protocol upgrade, but from a Chinese AI startup named Kimi. News broke that Kimi was facing a severe computing power shortage—its models were hitting the performance ceiling not because of code, but because of silicon. In the same 24-hour window, IREN, Hut 8, and a handful of former Bitcoin mining companies announced new AI cloud and data center contracts. Their stocks surged. Cipher Mining jumped 16.76%. Hut 8 climbed 10.45%. IREN rose 19.69%. CoreWeave and Nebius, the pure-play AI cloud providers, barely moved.
This was not a random market wobble. This was a narrative pivot—one that I’ve been tracking since my early days covering StarkWare’s privacy layers in 2017. Back then, the story was about zero-knowledge proofs and the math of secrets. Today, it’s about something far more tangible: the physical infrastructure that makes AI dreams run. The crypto mining industry, once the pariah of energy consumption, is quietly becoming the backbone of the AI economy. But as with every narrative in this space, the surface level is only half the truth. Yield wasn't the only thing being mined underground.
The Context: From Bitcoin to B200s
To understand why a Bitcoin mining company signing an AI cloud contract matters, you have to understand the asset swap. Bitcoin miners are, at their core, operators of specialized data centers. They have land, power purchase agreements, cooling systems, and the logistical expertise to run thousands of machines 24/7. When the 2022 bear market hit and Bitcoin’s hashprice collapsed, many miners were left with stranded energy assets and a desperate need for revenue diversification.
Enter AI. The compute demands of large language models and generative AI created a new market for high-performance computing (HPC) services. The same GPUs that could validate Ethereum transactions or mine cryptocurrencies could also train neural networks. But more importantly, the real estate and power infrastructure that miners had built for ASICs could be retrofitted for GPU clusters. The pivot was not just logical—it was inevitable.
IREN (formerly Iris Energy) is the poster child. The company operates hydro-powered mining facilities in British Columbia and Texas. In recent months, it secured AI cloud contracts with Microsoft, Nvidia, Perplexity, and Figure. The annualized recurring revenue target was revised upward to $4 billion. Hut 8, another major miner, signed a 15-year, $9.8 billion AI data center lease agreement with an unnamed hyperscaler. Cipher Mining and CleanSpark also announced expansions. The market’s reaction was euphoric.
But I’ve been here before. I watched the NFT art bubble inflate and deflate. I saw the LUNA collapse erase $40 billion in 72 hours. I learned that the most dangerous moment in a narrative is when it becomes too easy to believe. The question is not whether AI infrastructure is in demand—it is, and will be for years. The question is whether these miners are the right vessels for that demand.
The Core: Narrative Mechanism and Sentiment Analysis
The narrative at play is one of scarcity and substitution. The “Kimi computing power shortage” is not an isolated incident—it’s a microcosm of a global bottleneck. AI startups everywhere are hitting the same wall: they need more GPUs than the market can supply, and they need them now. Traditional cloud providers (AWS, Azure, GCP) are too expensive for small players and often lock customers into proprietary ecosystems. Enter the independent AI cloud providers—CoreWeave, Applied Digital, and now the Bitcoin miners.
The market is buying this story because it’s simple: AI needs compute, miners have compute, so miners will win. The stock movements confirm it. But sentiment analysis shows a gap between the bullish narrative and the underlying fundamentals. For instance, CoreWeave—the most established independent AI cloud provider—did not experience the same lift. This suggests that the market is not rewarding the sector uniformly, but rather picking stocks based on perceived “new news” (the contracts) rather than on sustainable business models.
Let’s look at the numbers. Hut 8’s $9.8 billion contract over 15 years implies an annual revenue contribution of roughly $653 million. That’s impressive, but it doesn’t account for the capital expenditure required to build the data center, the power costs (which could rise), or the risk of technological obsolescence. If the hyperscaler client decides five years from now that the GPU generation in that data center is obsolete, the contract may be renegotiated or terminated. The contract value represents total potential revenue over 15 years, not guaranteed profit.
The same caution applies to IREN’s $4 billion annualized recurring revenue target. How much of that revenue is locked in via contracts, and how much is projected based on capacity utilization? In my experience auditing crypto mining firms, targets are often aspirational. The real test will come when the next quarterly report reveals actual revenue and gross margins.
Yet, the sentiment is undeniably bullish. The narrative has legs because it taps into a deeper need: the fear of missing out on the AI boom. Investors who missed Nvidia’s run are looking for secondary exposure. Bitcoin miners offer that—with a crypto twist. It’s a story of redemption: from energy vampires to AI saviors. But as I wrote in “The Female Face of DeFi” back in 2020, redemption arcs are compelling but rarely complete.
The Contrarian: What the Narrative Overlooks
Here’s the contrarian angle, and it’s one that I’ve developed through years of covering both DeFi and AI. The thesis that Bitcoin miners will become dominant AI infrastructure providers has at least three blind spots.
First, the chip dependency. These miners are building data centers around Nvidia’s H100 and B200 GPUs. But Nvidia itself is a client and a competitor. By signing contracts with IREN, Nvidia is offloading some of its own compute demand, but it also has the option to build its own data centers. The same goes for Microsoft. The hyperscalers are using miners as a short-term bridge while scaling their own internal capacity. Once their own buildouts are online, they may reduce reliance on third parties. The “scarcity” narrative could flip to “overcapacity” within 18 months.
Second, the power constraint. The most valuable asset these miners have is their power purchase agreements. But as more data centers come online, demand for electricity is skyrocketing. In Texas, ERCOT has warned about grid strain. In Canada, hydro capacity is finite. The miners that were once lauded for using “stranded energy” may find that their power is no longer cheap or abundant. I recall a conversation with a mining executive in 2023 who said, “We thought electricity was our moat. Turns out, it’s our risk.”
Third, the lack of ecosystem lock-in. Unlike AWS or Azure, these miners offer bare-metal compute without the software stack. That means customer stickiness is low. AI companies can pack up their models and move to another provider as soon as a cheaper option appears. In a market where compute is fungible, margins will compress. We’ve seen this in crypto mining itself: after the Shanghai upgrade, ETH staking pools competed on fees, and the big players squeezed out the small ones. The same dynamic will play out in AI cloud.
And let’s not forget the lesson from the 2021 NFT bubble: when liquidity dries up and hype fades, floor prices collapse. The “blue chip” label becomes worthless. In that light, the “AI infrastructure” label is the new blue chip. But every narrative has a half-life.
The Takeaway: Where the Next Narrative Pivot Lies
So, what comes after this pivot? If I’ve learned anything from covering this space for eight years, it’s that the next narrative is already forming in the shadows. The convergence of AI and crypto is not just about compute—it’s about identity, trust, and provenance. AI-generated content is flooding the internet, and we have no reliable way to distinguish what is real. Decentralized identity protocols (like those using zero-knowledge proofs) could become the verification layer for AI output.
I’m currently spearheading a research collective in Tel Aviv that explores this very intersection. We’re analyzing how decentralized identity can authenticate AI-generated content without relying on centralized gatekeepers. It’s early, but the signal is growing. IREN and Hut 8 are building the pipes. The next wave will be about who controls the water.
For now, the market is celebrating the miners’ pivot. But I’d caution against blind optimism. The contracts are real, but so are the risks. The narrative is compelling, but it’s not the whole story. In the bear market of 2022, I learned that survival matters more than gains. Right now, these companies are surviving by pivoting. But true resilience will come when they build something that lasts beyond the next compute cycle.
As I wrote in “The Truth Protocol” earlier this year: Yield wasn’t the point. Trust is. And in the AI era, compute is just the beginning.