Hook
Forty-one point nine million dollars. That is what Core Scientific paid to walk away from a contract for Block’s 3-nanometer mining chips. Not because the chips were defective. Not because of a market crash. They paid to abandon a technology that Jack Dorsey had touted as the future of decentralized infrastructure. Then, almost immediately, they signed a fifteen-year deal with AMD to rent out their data center for AI workloads. The noise of the network—the constant hum of hashrate and hype—suddenly had a new frequency. This was not a story about a failed product. It was a story about a failed narrative.
Context
Block, formerly Square, entered the mining chip market with the ambition of breaking Bitmain’s oligopoly. Jack Dorsey’s vision was clear: a vertically integrated, open-source mining stack that would democratize Bitcoin’s security. The 3nm chip project, codenamed Proto, was supposed to be the hardware engine of that vision. Core Scientific, one of the largest publicly traded Bitcoin miners, was the anchor customer. They ordered enough chips to generate 15 exahash of computing power. It seemed like a marriage of narrative and capacity: Block’s cypherpunk ethos paired with Core’s industrial-scale mining. But within a year of the announcement, Core filed a notice of termination, taking a $41.9 million impairment charge. The same quarter, they announced a pivot to AI—leasing their Austin, Texas facility to AMD for high-performance computing. To understand why, you have to look beyond the press releases and into the energy economics of the network.
Core
Let’s talk about the chip itself. A 3nm ASIC miner is state-of-the-art in terms of process node. But in Bitcoin mining, the only metric that matters is joules per terahash. Block never published independent benchmarks for their chip. The only public data point was a target of 15 EH/s for Core’s order—but not the efficiency. In my years auditing smart contracts and reviewing hardware specs, I have learned that the hardest thing to audit is a promise. Block promised a better chip, but the code—the actual performance—couldn’t deliver. Core’s decision to eat a $41.9 million loss rather than deploy those chips speaks volumes. If those machines had been competitive with Bitmain’s S21 or MicroBT’s M60 series, would any rational operator walk away? No. They would have sold the hashrate on the open market. Instead, Core chose to pay a premium to not use them. That is the technical verdict: the chip’s real-world efficiency likely fell short of what was needed to generate profit in a post-halving environment where the network hashrate continues to climb.
But the deeper layer is narrative. The narrative is the asset; the code is the proof. And the proof here is that a major miner preferred to burn $41.9 million and pivot to AI than to double down on Bitcoin mining. This is not just a company decision—it is a signal about the relative economic gravity of two ecosystems. Bitcoin mining is a commodity business with thin margins, high capital expenditure, and relentless difficulty adjustments. AI compute, on the other hand, has customers willing to sign 15-year contracts at premium rates. Core’s deal with AMD is reportedly worth $14 billion in potential revenue over its term. Compare that to mining, where the same facility might generate a fraction of that, with volatility in both Bitcoin price and network difficulty. The market sentiment has shifted from 'hashing power is digital gold' to 'compute power is the new oil.' This is the core insight: Bitcoin mining narratives are being subsumed by AI narratives because the economic fundamentals point in that direction.
I recall a conversation with a mining operator in 2023. He told me, 'We’re not miners. We’re energy arbitrageurs with a Bitcoin hedge.' That cynicism was prescient. Core’s pivot is the logical conclusion of that mindset. They saw that their land, power contracts, and cooling infrastructure were worth more when rented to AI than when used to mine Bitcoin. The chip cancellation was simply the trigger. Block, meanwhile, is left holding a product that no one wants. Their other crypto initiatives—Tidal, TBD, Bitkey, Bitchat—have all failed or been shut down. Where code meets culture, the real value emerges, but Block’s code never met the culture of efficient mining. They built for a world where decentralization matters more than economics, but the market chose economics.
Contrarian
Now the contrarian angle: Core Scientific’s $41.9 million penalty was actually a brilliant strategic move. In traditional finance, a breakup fee is often seen as a cost of doing business when a better opportunity emerges. Core paid $41.9 million to unlock a $14 billion revenue stream. That is a return on investment that most hedge funds envy. The contrarian narrative is that Block’s failure is not a failure of Bitcoin or mining—it is a failure of execution by a company that overestimated its hardware capabilities. Bitcoin mining remains a viable business for those with the best chips and lowest power costs. Block simply didn’t have a competitive product.
Furthermore, the AI pivot is not without risk. The AI data center market is already crowded. Core’s deal with AMD could face execution delays, technology shifts, or a downturn in AI investment. Searching for truth in the noise of the network, I see a second-order contrarian idea: the exodus of mining capacity to AI might actually strengthen Bitcoin’s security in the long run. If only the most efficient miners survive, the network’s hashrate will be concentrated among operators with the best cost structures and the most to lose from dishonest behavior. That could make the network more secure, not less. The narrative of 'AI is stealing mining's resources' may be replaced by 'AI is forcing mining to become leaner and meaner.' But that is a story for later.
Takeaway
The next narrative is not about mining chips or even about mining itself. It’s about the grid—who controls the energy and compute resources that power both Bitcoin and AI. The war is now between two hungry ecosystems for the same atoms: electricity, land, and capital. For now, AI is winning the bidding war, as evidenced by Core’s willingness to pay $41.9 million to switch horses. But narratives evolve. Bitcoin’s fixed supply and global settlement utility are unique value propositions that no AI token can replicate. The question is whether those propositions can generate enough economic return to compete for energy resources in a world obsessed with artificial intelligence. I’ll be watching the next quarterly filings from Core, Marathon, and Riot. The signal is in the numbers, not the tweets. And the truth? It’s hiding in the noise of the network.