We didn’t see this coming, but the signs were already carved into the Q2 earnings release.

Core Scientific just paid $41.9 million in cash to tear up a contract for Block’s 3nm mining rigs. Not a renegotiation. Not a delay. A termination. Voluntary. Costly. Definitive.

Let that number sit: $41.9 million. That’s roughly 1,860 Bitcoin at current prices. They burned that to walk away from Jack Dorsey’s “revolutionary” Proto chip. Why?

Because staying would have cost them more. In capital, in opportunity, and in relevance.

Context: The Chip That Never Was

Block’s Proto project was launched in 2023 with typical Dorsey hype — open-source mining, decentralized silicon, a middle finger to Bitmain’s monopoly. The 3nm chip was supposed to deliver 15 Exahash of compute. The only public customer? Core Scientific. The only real test? A live deployment that clearly failed.

Now core is pivoting hard: from Bitcoin hashrate to AI compute. They signed a 15-year deal with AMD to lease their data centers, projected to generate $14 billion in revenue. That’s the new gold rush, and Core just sold their mining picks to buy a shovel for it.

Core: The Anatomy of the Breakup

The termination clause in the original contract gave Core an out — but at a steep price. Paying $41.9 million is a signal, not a calculation. Here’s what it tells us:

1. The chip’s performance was underwhelming. No independent benchmarks. No power efficiency numbers. Core’s own financial model likely showed negative ROI at current Bitcoin prices and network difficulty. Why mine at 20 J/TH when Bitmain’s S21 XP does 13? The gap isn’t marginal; it’s existential.

1. Core’s strategic pivot is real. In my years analyzing mining operations — from the 2017 ICO frenzy to the 2022 collapse — I have never seen a major miner voluntarily cripple their hashrate growth to enter a completely different business. Core is betting that AI compute will yield higher margins and less volatility than mining. They are essentially releasing their most capital-intensive assets (the new rigs) to reinvest in GPU capacity and power contracts.

1. Block’s mining roadmap is now a zombie. The only institutional buyer is gone. Block will have to write down the R&D, sell the chips at a discount on the secondary market, or shutter the division entirely. The “healthy pipeline” that Dorsey touted last quarter? A ghost.

But the deeper story isn’t just about Block’s failure. It’s about the structural reallocation of resources out of Bitcoin mining and into AI.

Contrarian: The Real Story Is Not Block’s Failure — It’s Mining’s Loss

The market will frame this as: “Block’s mining chip flopped; another Dorsey fantasy dies.” That’s true but trivial. The provocatively overlooked angle is the signal that Core Scientific — one of the largest public miners — has effectively declared that Bitcoin mining is a less attractive business than renting out the same power and land to an AI hyperscaler.

Think about the implications for Bitcoin’s security budget:

• Miners are the backbone of proof-of-work. If the most sophisticated operators choose to redirect their energy toward AI instead of hashing, the network’s long-term hash rate growth slows.
• This isn’t a temporary cycle dip. It’s a permanent shift in the cost of capital. AI companies pay premium rates for high-density compute. Miners negotiate thin margins on volatile Bitcoin prices. The rational choice is obvious.
• The narrative that “miners can always sell power back to the grid” is outdated. Now they can sell it to AI — and they will.

Core’s $41.9 million goodbye is the opening bid in a larger auction. Expect more miners to follow. Marathon, Riot, even privately held operations will start evaluating whether their substations and cooling infrastructure are better suited for NVIDIA H100s than S19s.

This is the death of the “pure-play bitcoin miner” as a viable long-term strategy. The evolution of the industry is happening right now, in plain sight, hidden inside a quarterly footnote.

Takeaway: Where the Next Hash War Will Be Fought

So ask yourself: If the largest miners become AI landlords, who secures Bitcoin’s transaction finality?

The answer might be “no one” — or rather, a thinner, less robust set of dedicated hobbyists and industrial laggards. The next Bitcoin halving will reveal whether the security model depends on mining profitability or on something more fragile: inertia.

For now, watch Core Scientific’s next two quarterly reports. If AI revenue surpasses mining revenue — and it will — the paradigm has shifted. Block’s 3nm chip will become a footnote, but the real history being written is the slow, quiet reallocation of energy away from Bitcoin’s proof-of-work.

We didn’t see the fire until it was already burning. The smoke is $41.9 million thick.

Based on my experience auditing the capital structure of mining operations during the 2022 bear market, I can tell you: writing off a contract like this is never a gamble. It’s a calculation. And Core Scientific just calculated that Bitcoin mining’s return on energy is lower than AI’s. That arithmetic changes everything.