The Great Digital Value Divergence: AI Commoditization and Bitcoin Scarcity

Arthur Hayes' recent remarks have stirred both the tech and investment communities: generative AI models (such as OpenAI and Anthropic) may lack long-term intrinsic value, while Bitcoin is the ultimate winner. If you dismiss this as mere technological skepticism, you miss its real significance. From the perspective of IT governance and financial structure analysis, this judgment points to a fundamental split in the digital economy—"information utility" and "value storage" are decoupling. What we face is not one replacing the other, but the formation of a two-tier digital architecture: one layer is liquid commoditized intelligence, the other is stock-like immutable scarcity.

The Structural Asymmetry Between Flow-Type Technology and Stock-Type Assets

To understand the investment implications, we first need to see the economic structures behind these two types of technology. Large language models are "flow-type" technology: their utility is tied to continuous input—massive compute, energy, and constantly refreshed data. AI's economic value is anchored in runtime performance and information-processing capability, yet it is therefore structurally dependent on centralized GPU clusters and cloud infrastructure. This dependence gives rise to "intelligence monopolies" while bringing regulatory and operational risks.

The direction of competition is key. As performance gaps between models narrow, the "commoditization of intelligence" is almost inevitable: AI is shifting from a differentiating competitive factor to a standard utility, becoming part of digital infrastructure. This does not mean AI has no value, but that its value takes the form of high turnover, low margins, and high substitutability.

Bitcoin's logic is entirely different. Through proof of work (PoW), it converts physical energy into mathematical scarcity. AI models require continuous updates and are subject to the will of centralized providers, whereas Bitcoin's protocol rules are mathematically fixed and decentralized. AI provides the digital age's "processing capability"; Bitcoin provides the "settlement layer." Bitcoin's value comes not from the utility of processing data, but from resistance to modification and a mathematically capped supply.

Proof of Work (PoW) is Bitcoin's consensus mechanism: miners compete for the right to record transactions by repeatedly computing hash values, a process that consumes real electricity and hardware compute and whose results can be verified at low cost by the entire network. Its economic significance lies in the fact that attacking the network requires controlling more than half of the hashing power, and the cost rises with the network's scale, so security itself becomes a continuously invested "sunk cost." Compared with trust models that rely on centralized servers to issue credentials, PoW outsources trust to physical energy and mathematical rules rather than to any institution's reputation. This is precisely the basis for "converting physical energy into mathematical scarcity" mentioned in the text, and it also explains why Bitcoin's supply cap (21 million coins) can be institutionally locked in without any central institution's promise.

The Emergence of a Two-Tier Digital Economy

For global investors, the key point is that digital markets are reorganizing into two tiers.

The Intelligence Layer: High Speed, High Utility, Low Margin

This layer is driven by AI agents and models, delivering unprecedented operational efficiency and automation. But it depends on centralized capital and faces "value uncertainty"—bias, regulatory changes, and service disruptions can all hit its stability. Therefore, this layer cannot serve as a reliable store of value. Its role is the engine, not the vault.

The "commoditization of intelligence" refers to AI capabilities evolving from differentiated products of a few vendors into interchangeable, metered, standardized services. Its drivers include: open-weight models continually catching up with closed-source models, inference costs falling rapidly by generation, and multiple vendors offering APIs with similar functionality. The economic consequence is similar to utilities such as electricity and bandwidth—once service quality converges, competition centers mainly on price and reliability, making it hard for vendors to sustain high margins. For users, this means "intelligence" shifts from a strategic asset to a procurable input; for investors, it means a single model vendor's moat may be shallower than expected, and its valuation logic is closer to that of an infrastructure operator than a monopoly platform.

The Value Layer: A Neutral, Permissionless, Tamper-Proof Settlement Mechanism

As autonomous transactions between AI agents rise, the demand for neutral, permissionless, tamper-proof settlement mechanisms will become urgent. Bitcoin's positioning is precisely the global decentralized settlement layer of this automated economy. In a world flooded with AI-generated content and synthetic data, the need for a non-custodial, immutable value ledger will shift from "optional" to "economic necessity."

These two layers are not a zero-sum relationship. AI provides the cognitive engine; Bitcoin provides the value foundation. The real opportunity lies in recognizing that intelligence is becoming a commodity, while scarcity is becoming the ultimate premium.

"Permissionless" means any entity can read the ledger, initiate transactions, or participate in validation without prior institutional approval. This contrasts with the access-control systems of traditional finance: cross-border settlement often requires multiple authorizations from banks, clearinghouses, and regulators. For high-frequency, small-value autonomous transactions between AI agents, permissionless means programmatic wallets can complete payments plug-and-play without signing commercial agreements one by one or going through manual approval. At the same time, "non-custodial" means private keys are held by the owner, and the assets do not constitute liabilities of a custodian, thereby avoiding counterparty default risk. Together, these two form the feasibility premise for what the text calls the "settlement layer" of the automated economy.

IT Governance and Corporate Treasury Response Strategies

From the perspective of IT governance and corporate treasury, the convergence of these two layers requires proactive strategic adjustment. Organizations need to move beyond thinking of AI and blockchain as separate silos and incorporate them into a unified digital risk and capital management framework.

Mitigating Technology Dependence

Enterprises should adopt a "multi-model" AI strategy to avoid vendor lock-in. Relying on a single centralized AI provider poses systemic risk to business continuity. Governance frameworks should prioritize interoperability and resilience, keeping the intelligence layer as a modular utility rather than a single point of failure.

Ensuring Data Provenance

In an era where AI-generated content is everywhere, the "authenticity" of data becomes a core competitive advantage. We recommend incorporating blockchain-based traceability into IT governance to ensure the integrity and ownership of enterprise data and to resist the erosion of facts by synthetic information.

Integrating Digital Asset Management

As AI-driven automation generates surplus capital through efficiency gains, the treasury function must also evolve. The integration of IT and financial strategy should include incorporating decentralized, inflation-resistant digital assets into long-term capital allocation strategies.

Divergence, Not Replacement

The digital economy is not a zero-sum game between AI and Bitcoin, but a structural divergence. AI will become the cognitive engine of the future, while Bitcoin constitutes the immutable foundation of value. For sophisticated investors, seeing this is enough: intelligence is commoditizing, and scarcity is becoming the ultimate premium.

One-line Summary

AI is moving toward commoditized intelligence while Bitcoin carries scarce settlement—the digital economy is heading toward a two-tier divergence.

Paragraph Summary

Starting from Arthur Hayes' judgment that "generative AI lacks long-term intrinsic value and Bitcoin is the winner," the article argues that this is not technological skepticism but a revelation of the decoupling of "information utility" and "value storage" at the foundation of the digital economy. Large language models are flow-type technology whose utility depends on continuous inputs of compute, energy, and data, and is constrained by centralized infrastructure; as model gaps narrow, intelligence will commoditize into high-turnover, low-margin, highly substitutable services. Bitcoin, through proof of work, converts energy into mathematical scarcity, with value derived from resistance to modification and a fixed supply cap, positioning it as a settlement layer. Together they form a two-tier architecture: the intelligence layer is the engine, the value layer is the vault. The article further offers corporate governance recommendations, including multi-model AI strategies, blockchain data provenance, and incorporating digital assets into treasury allocation, emphasizing that the two represent a structural division of labor rather than zero-sum replacement.