The discussion about the relationship between artificial intelligence and crypto has stopped being a speculative exercise about technological convergence.
In the current quarter, the relevant variable is material: both industries compete for scarce inputs, and one of the two has the capacity to pay prices the other cannot match.
The position defended in the article is that the crypto sector has underestimated the second-order effects of the asymmetry, and the impact is not limited to Bitcoin mining or AI-linked tokens.
Physical Overlap Is the Starting Point
The overlap between Bitcoin mining and AI data centers is not conceptual; it is engineering. Both activities require high-capacity electrical interconnection, cooling systems with high thermal density, rack space, and personnel specialized in continuous operations.
An industrial mining site and a training cluster share most of the technical requirement list. The difference is in the return.
Figures published by operators and sector contracts place revenue per megawatt for Bitcoin mining in a range close to one million dollars annually.
AI-oriented computation, measured through infrastructure contracts signed in the last eighteen months, sits at a scale ten to twenty times higher.
The gap does not admit interpretation: any board with fiduciary obligations to shareholders shifts capacity toward the use with higher return per unit of energy.
Contracts Are Already Signed
The displacement is not hypothetical. Core Scientific formalized an agreement with CoreWeave for more than ten billion dollars.
TeraWulf accumulates revenue commitments from HPC infrastructure above twelve billion. Hut 8 signed an AI infrastructure lease for fifteen years at seven billion.
The aggregate sum of announced contracts in the listed mining sector exceeds seventy billion dollars, and revenue mix projections point to most listed companies obtaining close to seventy percent of their revenue from AI activities by the end of 2026.
The correct reading of the process is not that miners “abandon” Bitcoin, but that they reconfigure their risk profile. A company that previously was a leveraged proxy to the underlying asset price begins to behave as an infrastructure landlord with long-term contractual flows.
Competition for Capital Is the Main Channel
The second vector is capital allocation. Hyperscaler capex budgets for AI infrastructure sit in the range of 650 to 700 billion dollars annually.
No inflow cycle toward digital assets recorded to date approaches the magnitude. The opportunity cost for a fund with discretionary mandate is explicit: every dollar allocated to a digital asset with double-digit volatility competes against an infrastructure thesis with signed contracts and verifiable demand.
The competition operates independently of the technical quality of protocols. It is not resolved with improvements in scalability or fee reductions.
It is resolved when capital perceives a risk-adjusted return differential that justifies the allocation, and in the current quarter the differential favors computation.
The Transmission Channel Is Equity Markets
The immediate risk does not come from competition for energy, but from correlation with equity markets. AI concentrates a growing portion of U.S. equity market capitalization. Any episode of risk aversion originating in valuations, regulation, or doubts about investment returns propagates quickly toward digital assets.
The fourth quarter of 2025 offers empirical evidence. Bitcoin moved from highs near 126,000 dollars to a low below 86,000, with a quarterly return among the worst recorded after 2022.
The trigger was not a crypto-native event, but a discussion about adversarial governance mechanisms for autonomous agents, raised from within the ecosystem, which coincided with a broader debate in the technology sector about regulatory controls, upward pressure on energy prices, a Federal Reserve with a restrictive bias, and political challenges to AI industry requirements.
The operational conclusion is uncomfortable: crypto has no representation at the tables where AI policy is decided, but its price is formed partially in function of the decisions.
Supply Pressure from Corporate Treasuries
A third channel exists, less discussed. The transition toward AI requires intensive capex, and listed mining companies need to finance it.
BTC treasuries constituted a strategic reserve when the main business was mining; they lose justification when the business becomes contracted infrastructure.
Aggregate holdings reduction in the sector exceeds fifteen thousand BTC from highs, with cases of practically total liquidation of remaining positions.
The aggregate balance of Bitcoin on listed miner balance sheets approaches 80,000 BTC. The magnitude is not catastrophic in terms of market depth, but the direction of flow is relevant: supply appears in a context of restricted liquidity and post-reset deleveraging. A seller with a defined calendar and no strategic motivation to retain position exerts pressure on the order book.
The Convergence Thesis Does Not Solve the Liquidity Problem
The usual counterargument holds that AI agents will need settlement rails and blockchain protocols will occupy the function. The thesis is plausible over a multi-year horizon. Permissionless settlement, 24/7 operating markets, and verifiable computation are capabilities that a financial system populated by automated agents could require.
However, a five-year adoption thesis does not alter portfolio allocation for the quarter. The market discounts flows, not possible architectures.
Presenting convergence as a response to the restricted liquidity environment constitutes a time-horizon error, and in my reading it is the type of reasoning that has prevented the sector from recognizing the magnitude of the capital displacement in course.
What the Sector Should Monitor
The adequate response is not rhetorical. It is operational. Monitoring should concentrate on four variables: hyperscaler capex announcements and their financing, new contracts for conversion of mining capacity toward HPC, quarterly disclosures of corporate treasuries in BTC, and evolution of correlation between digital assets and technology indices.
A fifth element deserves attention: the capacity of the crypto sector to articulate concrete proposals on governance of autonomous agents before regulation is drafted without its participation.
The discussion about AI controls advances in forums where the digital asset industry has no seat. Incorporating late implies operating under rules designed for another set of actors.
The conflict between AI and crypto is not a war of narratives or a competition between technical communities. It is a dispute over capital, energy, and physical infrastructure, with price asymmetries that the crypto sector cannot correct through discourse.
Mining has already partially migrated. Capital has already partially migrated. Correlation with equity markets has already materialized in price data.
The correct bias for the quarter is caution, not because of structural weakness in protocols, but because of verifiable competition for the same resources.
The alternative, insisting on technological convergence as an allocation argument, reproduces the pattern which already produced the most severe position adjustment since 2022.
