A Response to Crespo's Article That AI Is Killing Bitcoin
by Alexandra Indra Kruse for this Carl Kruse Blog
The article "Anthropic is Killing Bitcoin" by Jose Crespo
(published on Medium/AI Advances) presents a provocative thesis: that the rise
of Artificial Intelligence—specifically companies like Anthropic—will lead to
the "death" of Bitcoin as miners migrate to AI data centers and
"AI tokens" replace the monetary function of BTC.
While the pivot of some mining firms toward AI is a real trend, the
argument that this "kills" Bitcoin overlooks fundamental technical,
economic, and game-theoretical realities of the Bitcoin network. Below are the
primary counter-arguments to this thesis.
1. The Hardware Incompatibility Fallacy (ASICs vs. GPUs)
The claim that miners can simply "abandon" Bitcoin mining to serve as AI data centers ignores a massive technical hurdle: Hardware specialization.
- ASICs are Not GPUs: Bitcoin mining is performed by ASICs (Application-Specific Integrated Circuits) designed solely to compute SHA-256 hashes. These machines cannot run AI models, train LLMs, or perform inference. They are literal "bricks" in an AI context.
- Total Re-Investment: For a miner to pivot to AI, they must scrap their entire fleet of mining hardware and purchase GPUs (like NVIDIA H100s), which are currently in short supply and cost tens of thousands of dollars each. Most "pivoting" firms are not repurposing hardware; they are merely using their existing land and power contracts to build entirely new, separate business lines.
2. Bitcoin’s "Difficulty Adjustment" ensures survival
The article implies that if miners leave, the network becomes vulnerable or dies. This ignores Bitcoin’s most elegant feature: the Difficulty Adjustment.
- Self-Balancing: If 50% of miners leave the network today, the difficulty of mining drops by 50% in the next adjustment period. This makes mining twice as profitable for those who remain.
- Market Equilibrium: As long as Bitcoin has any market value, there will always be an economic incentive to mine it. The exit of "weak" or "distracted" miners actually improves the profit margins for dedicated miners.
3. Utility vs. Store of Value (Commodity vs. Money)
The author suggests "AI tokens" (referring to the units of text/computation used by LLMs) will replace BTC’s monetary function. This conflates utility commodities with monetary assets.
- Consumables vs. Stores of Value: An AI token is a "consumable"—you spend it to get a result (an answer from Claude or GPT). Its value is tied to the cost of compute and electricity.
- Infinite Supply: As compute becomes more efficient (Moore's Law), the "supply" of AI tokens effectively increases. Money requires scarcity. Bitcoin’s 21-million-unit cap makes it a "hard" monetary asset. AI tokens are "soft" commodities that will likely trend toward a marginal cost of zero over time.
4. Differing Energy Profiles
The article assumes AI and Bitcoin are in a zero-sum war for electricity. However, their energy "appetites" are fundamentally different:
- AI Needs "High-Quality" Power: AI data centers require 99.99% uptime, low latency, and high-speed fiber optics. They must be located near urban hubs or stable grids.
- Bitcoin Thrives on "Stranded" Power: Bitcoin miners are "energy buyers of last resort." They can operate in the middle of a desert using flared natural gas or in remote mountains using excess hydro power that cannot be sent to cities. Because they can be turned off instantly (interruptible load), they complement renewable energy grids in ways AI data centers (which must stay on) cannot.
5. Bitcoin as the "Native Currency" for AI Agents
Rather than AI killing Bitcoin, many experts (including CEOs like Jeremy Allaire and Arthur Hayes) argue that Bitcoin is the perfect currency for AI.
- Neutral & Permissionless: AI agents cannot open bank accounts or hold credit cards. They need a digital-native, permissionless protocol to pay for resources.
- The Lightning Network: AI agents can use the Lightning Network to send sub-cent micropayments for API calls or data. In this scenario, AI doesn't replace Bitcoin; it becomes Bitcoin's largest "user base," potentially driving demand for BTC to new heights.
6. Institutional Lindy Effect
Bitcoin has a 15-year head start in establishing "monetary Lindy" (the idea that the longer something survives, the longer it is likely to persist).
- Regulatory Clarity: Bitcoin is the only digital asset classified clearly as a commodity by the SEC and CFTC.
- Trillions in Infrastructure: With the launch of Spot ETFs (BlackRock, Fidelity), Bitcoin is now integrated into the global financial plumbing. "AI tokens" or "compute credits" currently lack the legal framework, custody solutions, and institutional trust required to serve as a global reserve asset or monetary standard.
Conclusion
The "Anthropic is Killing Bitcoin" argument is a classic
example of category error. It treats Bitcoin as a tech startup that can
be "disrupted" by a better app. In reality, Bitcoin is a monetary
protocol. Just as the invention of the airplane did not "kill"
the use of gold as a reserve asset, the invention of more efficient
"intelligence units" will not kill the world’s first decentralized,
scarce digital money. If anything, the AI revolution will increase the global
demand for a neutral, verifiable, and scarce asset like Bitcoin to settle value
between machines.
++++++++++++++++
The homepage for this Carl Kruse Blog is at
https://carlkruseofficial.blogspot.com
Contact: carl At carlkruse DOT com
Other articles in the blog include: The Culture of Gender, Planning Makes Your Life Happier, and Retrofitting Classic Cars.
Another Carl Kruse Blog is at
https://carlkruseofficial.wordpress.com/

Comments
Post a Comment