Core Scientific just paid $41.9 million to break a contract for Block’s 3nm mining chips. That’s not a loss—it’s a strategic signal. The largest publicly traded Bitcoin miner in North America is walking away from hardware it ordered less than a year ago. They’d rather forfeit the cash than deploy Jack Dorsey’s proto ASICs. Why? Because the narrative around Bitcoin mining has shifted—and Block’s chip failed to keep pace.
Context: The Rise and Fall of Block’s Silicon Dream
In early 2025, Block—the payments company formerly known as Square—announced its entry into the Bitcoin mining chip market with a 3nm ASIC promising 15 exahash per second. CEO Jack Dorsey framed it as a mission to decentralize mining hardware, challenging Bitmain’s near-monopoly. Core Scientific, the largest US-based mining host, signed on as the inaugural customer, ordering a significant batch.
But by Q4 2025, Core Scientific disclosed a $41.9 million impairment charge related to terminating that contract. The official reason: “strategic pivot” toward AI data center hosting. The same week, they signed a 15-year deal with AMD to lease 500 megawatts of power for AI compute—a contract estimated to generate $14 billion in revenue. The message was clear: Bitcoin mining no longer offers the risk-adjusted returns that AI infrastructure does.
Core: Why the Chip Failed and What Broke the Narrative
From a technical standpoint, Block’s 3nm chip was a legitimate engineering effort. But in the brutal commodity market of ASICs, performance is measured by one metric: energy efficiency, or J/TH. Block never published independent benchmarks, and the only public test—the deployment at Core Scientific—was aborted.
During my 2020 audit of Uniswap’s MEV risks, I learned that the gap between a prototype and a production-ready system is often lethal. In mining, that gap is even wider. Bitmain and MicroBT have spent a decade refining supply chains and optimizing thermal management. A new entrant like Block, even with Jack Dorsey’s brand and a 3nm process, cannot skip that learning curve. The chip may have achieved theoretical efficiency, but real-world stability and cost per hash likely fell short.
Furthermore, the timing was catastrophic. The 2025 halving had already compressed miner margins. Core Scientific—already navigating bankruptcy restructuring in 2023—could not afford to deploy underperforming hardware when competitors like Riot and Marathon were buying the latest S21 series from Bitmain. The $41.9 million termination fee was cheaper than the losses from running Block’s chips for two years.
Narrative is the new liquidity. The narrative that “Bitcoin mining is a digital gold rush” has been replaced by “AI compute is the new oil.” Core Scientific is simply following the capital. They are not abandoning Bitcoin—they are optimizing for returns.
Contrarian: Block’s Failure Might Be Good for Bitcoin Mining
Here’s the counter-intuitive angle: Block’s exit from the ASIC market could actually strengthen Bitcoin’s security model in the long run. The mining industry is consolidating around two dominant suppliers—Bitmain and MicroBT—which creates obvious centralization risks. But Block’s failure proves that hardware competition is not a public good; it’s a capital-intensive zero-sum game. New entrants waste billions chasing marginal efficiency gains.
Instead, the real innovation in mining is happening at the infrastructure level. Core Scientific’s pivot to AI hosting shows that miners can repurpose their power assets for dual use: selling compute to AI during low-difficulty periods and flexing back to Bitcoin during high-difficulty cycles. This hybrid model reduces the risk of hash rate volatility and makes mining more resilient.
The contrarian view is that Block’s chip retreat is a signal that the ASIC market has reached a natural oligopoly. Attempts to democratize it will fail unless backed by orders of magnitude more capital. Hype is cheap. Strategy is expensive.
Takeaway: What Every Investor Should Watch Next
Core Scientific’s decision is a canary in the coal mine for the entire mining sector. The next 12 months will see a wave of miners pivoting to AI or shutting down. Those who can negotiate long-term power contracts and land deals for AI data centers will survive. Those who stay purely focused on SHA-256 will face margin compression.
For Block, this is another entry in a growing list of failed crypto ventures: Tidal (music), TBD (Web5), Bitkey (self-custody wallet), and now Proto mining. Jack Dorsey’s vision of a decentralized financial stack is bleeding cash. The question is not whether Block will shut down Proto—but when.
The market is voting with its feet. Core Scientific walked away from $41.9 million worth of hardware to chase AI revenue. That vote is louder than any press release.