There is a moment in every technology cycle when the abstract promise collides with the physical ledger. For Jack Dorsey’s Block, that moment arrived not in a boardroom, but in a contract cancellation notice. Core Scientific—one of the largest Bitcoin mining hosts in North America—chose to pay $41.9 million in penalties rather than take delivery of Block’s 3nm Proto mining chips. The number is not a fine. It is a confession.
We chart the code, but the soul chooses the path. And in this case, the path led away from Dorsey’s hardware dream and toward a 15-year, $14 billion contract with AMD for AI compute. The story is not about a failed chip. It is about a structural realignment of value in the digital asset ecosystem—one that carries profound implications for how we measure decentralization, trust, and the future of Bitcoin’s security model.
Context
Block’s Proto chip was supposed to be a wedge into the oligopolistic Bitcoin mining hardware market, dominated by Bitmain and MicroBT. Dorsey framed it as a mission to decentralize mining—a narrative that resonated with the Cypherpunk ethos. The chip was built on TSMC’s 3nm process, a cutting-edge node that promised efficiency gains. Core Scientific, a mining giant that had emerged from Chapter 11 bankruptcy in early 2024, became the anchor customer, committing to a 15 EH/s purchase.
But between the announcement in late 2025 and the delivery window in early 2026, something shifted. Core Scientific’s management, led by a team that had already proven its adaptability during the crypto winter, recalculated. They saw that renting the same power and cooling infrastructure to AMD for AI training would generate more predictable, higher-margin revenue than burning electricity to mine Bitcoin at post-halving difficulty levels. The $41.9 million penalty was a cheap tuition fee for a lesson in opportunity cost.
This is not the first time I have seen idealism collide with physics. During my work translating Ethereum Classic’s “Code is Law” doctrine for Spanish-speaking communities in 2017, I learned that immutable principles only survive when the underlying incentives are aligned. When they aren’t, the code bends—or breaks.
Core
The Signal Behind the Settlement
The most technically relevant detail is not the 3nm process node; it is the fact that no independent benchmark of Block’s chip was ever published. In an industry where J/TH (joules per terahash) is the true currency, silence is data. Based on my auditing experience across L1 protocols and hardware supply chains, I have found that when a manufacturer refuses to release third-party power efficiency numbers, it is usually because those numbers are uncompetitive.
Consider the current market leaders. Bitmain’s Antminer S21 series achieves around 17 J/TH. MicroBT’s M60 series is comparable. If Block’s Proto chip had been even 10% better, we would have seen marketing blitzes. Instead, we saw a press release and a cancellation.
The core insight is this: the Bitcoin mining hardware market has reached a point of technological maturity where incremental improvements no longer justify switching costs. The network effects of Bitmain’s aftermarket service, spare parts availability, and resale liquidity create a moat that cannot be crossed by a new entrant with a slightly better chip.
This echoes a pattern I documented during the DeFi Summer of 2020. When MakerDAO’s overcollateralization model was challenged by flash loan attacks, the community chose to audit and harden existing code rather than adopt a radically new architecture. The same conservatism applies to miners: they will pay a premium for proven reliability.
Core Scientific’s Pivot as a Stress Test for Bitcoin’s Security Budget
Core Scientific’s move to sign a 15-year, $14 billion contract with AMD for AI/HPC compute is not merely a business pivot. It is a stress test of Bitcoin’s security budget. The Bitcoin network’s security is funded by block rewards and transaction fees, which ultimately come from the mining hash power’s profitability. If the marginal return on a watt of power is higher when allocated to AI training than to SHA-256 hashing, then rational miners will redirect their energy—and their capital.
This creates a downward pressure on Bitcoin’s equilibrium hash rate. The network adjusts difficulty downward when hash rate drops, but that is a lagging response. The real risk is a persistent diversion of new capital away from mining, which could leave the network more vulnerable to a concentrated attack by state-level adversaries who do not care about profit.
I lived through a similar structural shift in 2022, when I audited failing L1 protocols and found that the most robust chains were not those with the best technology, but those with the most resilient economic incentives. Bitcoin’s incentive model—miners paid in a volatile asset—has been its strength and its weakness. In a world where AI offers dollar-denominated fixed contracts, the weakness becomes more pronounced.
Block’s Broader Crypto Ambitions: A Pattern of Failure
The Proto chip cancellation is not an isolated incident. It is the latest in a string of failed crypto initiatives from Dorsey’s Block. Tidal (music streaming) and TBD (decentralized identity) have been shuttered or written down. Bitkey (self-custody wallet) has not achieved significant adoption. Cash App’s crypto trading faced regulatory scrutiny, resulting in over $200 million in fines. Block’s stock has fallen 68% over five years.
This pattern reveals a fundamental mismatch between corporate structure and decentralized ethos. Dorsey may genuinely believe in Bitcoin maximalism, but a publicly traded company accountable to shareholders must generate returns. Crypto projects require patient, mission-driven capital—not quarterly earnings pressure. The irony is that Dorsey, the ultimate free-speech advocate, is constrained by the very market mechanisms he seeks to disrupt.
I saw this tension firsthand when I collaborated on a Soul-Bound Token project for indigenous Mexican artists in 2021. The project succeeded because it was small, mission-aligned, and free from investor expectations. The moment we sought scale, the integrity frayed. Block’s failure is a case study in the limits of bringing corporate muscle to a community network.
Contrarian
The Counter-Intuitive Case: Why Block’s Failure Might Be Healthy
It would be easy to interpret this story as a bearish signal for Bitcoin mining or for decentralization. But there is a contrarian angle worth considering.
A more centralized mining chip supply—dominated by Bitmain and MicroBT—may actually be more predictable and auditable than a fragmented market. When everyone uses the same hardware, the network’s attack surface is known. When new players introduce untested chips, they may contain backdoors or vulnerabilities that are not discovered until it is too late. The Cypherpunk ideal of many small miners using diverse hardware is beautiful, but it is also a security nightmare. Verified uniformity can be safer than chaotic diversity.
Moreover, Core Scientific’s pivot to AI does not mean it is abandoning Bitcoin. It means it is diversifying its revenue streams, which may allow it to survive longer and continue contributing hash power during bear markets. A financially stable miner is better for Bitcoin security than a desperate one.
I recall a similar dynamic during the 2018 bear market. The miners that had hedged with other revenue sources—like hosting services—survived and thrived. Those that went all-in on Bitcoin alone perished. The market’s verdict on Core Scientific’s strategy is not a rejection of Bitcoin; it is a maturation of the industry.
The Blind Spot: AI Demand Might Also Be Cyclical
The contrarian must also question the permanence of AI demand. If the current AI boom is a bubble, then Core Scientific’s $14 billion contract is an albatross, not a lifeline. AMD might cancel or renegotiate if the market turns. The same volatility that affects Bitcoin could affect AI compute pricing. The assumption that AI is a stable, long-term revenue source is speculative.
I have seen this before in the ICO boom of 2017—projects pivoted to “blockchain for X” and then crashed when the hype faded. The lesson is that both Bitcoin mining and AI compute are subject to the same fundamental law: capital follows narrative, and narratives rotate.
Takeaway
The $41.9 million penalty paid by Core Scientific is a receipt for a lesson that will shape the next decade of digital infrastructure. Block’s chip failure and the pivot to AI are not endpoints; they are signals of a deeper transformation in how we allocate energy, capital, and trust.
We chart the code, but the soul chooses the path. The soul of Bitcoin mining is being asked to choose between a volatile, ideologically pure path and a diversified, economically pragmatic one. The answer is not obvious. But the decision is being made right now—in boardrooms, data centers, and on-chain transactions.
The question that haunts me is this: If the most rational economic actors redirect their resources away from Bitcoin’s security budget, what substitutes for that loss? A harder fork? A change in consensus rules? Or simply a more humble network, secured by hobbyists and true believers rather than corporations?
The path is not yet written. But the first exit sign has been illuminated—and it reads $41.9 million.