The numbers are stark. A CNBC and Generation Lab survey from August 13 shows 45% of Americans aged 18 to 34 expect AI to hurt their careers. Only 10% see it as a net positive. Bridgewater Associates estimates 18% of current US jobs could be displaced within five years. Andrew Yang, the former presidential candidate and now CEO of Noble Mobile, used these data points on CNBC’s Power Lunch to renew his push for an AI tax — a levy on companies that replace human labor with artificial intelligence, shifting the tax burden from payroll to algorithmic output.
Yang’s argument is structurally clean: if firms skip payroll taxes and healthcare costs by choosing AI over new hires, the government should tax the machine instead of the worker. He points to Dario Amodei, CEO of Anthropic, who floated a 3% AI revenue tax in 2025. The idea is simple — every time a model generates revenue, a fraction goes to the state. Yang proposes sending that revenue directly to displaced workers as checks, bypassing retraining programs he calls "largely failed." He cites coal miners and warehouse staff as examples.
Context: The Macro Liquidity Map
This debate is not new. Yang built his 2020 campaign on automation warnings, proposing a universal basic income called the Freedom Dividend. He also advocated for cryptocurrency adoption and clearer digital asset rules. Now, as a CBDC researcher who has spent years auditing liquidity cascades, I see a deeper structural issue: Yang’s AI tax is a regulatory response to a monetary phenomenon. The real problem isn’t taxation — it’s the collapse of labor-based income as a primary source of economic participation.
Consider the customer service sector. It employs roughly 2.9 million Americans, according to the Bureau of Labor Statistics. If 18% of US jobs are displaced within five years, that’s over 30 million workers. The tax base shrinks. The social safety net frays. Yang’s solution — a 3% revenue tax on AI — sounds precise, but it ignores how capital actually flows. AI models don’t generate revenue in a vacuum; they sit on top of compute infrastructure, data pipelines, and tokenized incentive layers. Taxing the output without taxing the inputs creates arbitrage opportunities.
Core: The Technical Fault Line
Let’s examine the proposed mechanism. A 3% AI revenue tax would apply each time a model generates revenue. That sounds like a simple point-of-sale levy, but in practice, it’s a nightmare to enforce. Revenue attribution is opaque. If a large language model powers a chatbot that leads to a sale, what fraction of that sale is "AI-generated revenue"? If the model is open-source and self-hosted, who is liable? The tax is designed for a centralized, auditable economy — precisely the kind of structure crypto was built to bypass.
This is where the liquidity cascade kicks in. Capital seeks the path of least resistance. If the US imposes a 3% AI revenue tax, firms will migrate their AI operations to jurisdictions with no such tax. They’ll move the model inference to a data center in Singapore or a decentralized compute network on a blockchain. The tax base evaporates. Yang’s proposal assumes the state can capture value from AI, but the machine economy is inherently global and borderless. Liquidity doesn’t lie. It will flow to where the regulatory friction is lowest.
Central banks are the ultimate VCs. They understand this better than most. The Federal Reserve, the ECB, and the People’s Bank of China are all exploring digital currencies partly to maintain monetary sovereignty in an era of programmable money. If AI starts generating value that is not captured by traditional payroll taxes, the state’s ability to fund itself erodes. A CBDC with programmable tax features — automatic withholding at the smart contract level — could be a more effective tool than a blunt revenue tax. But that requires a level of digital infrastructure the US currently lacks.
Contrarian: The Decoupling Thesis
Here is the contrarian angle: the AI tax debate is a distraction. The real decoupling is not between labor and capital, but between human labor and economic value generation. The Bridgewater estimate of 18% job displacement within five years is conservative. If you look at the trajectory of autonomous agents and machine-to-machine transactions, the percentage of GDP generated by human labor could drop below 50% within a decade. At that point, taxing AI revenue is like taxing the electricity that powers a factory — it’s a second-order effect, not a root cause.
Yang’s proposal to send tax revenue as direct checks is a UBI variant, but it misses the deeper issue: the structure of money itself. In a world where AI generates most value, the traditional monetary system — backed by labor and debt — becomes unstable. The protocol is the jurisdiction. The tax code is a legacy mainframe trying to regulate a quantum computer. Instead of debating how to tax AI, we should be debating how to redesign the monetary system so that value created by machines can be distributed to humans without breaking the state’s balance sheet.
Takeaway: Cycle Positioning
This is not a policy debate; it’s a liquidity signal. The fact that a former presidential candidate and a CEO of a major AI firm are both calling for an AI tax tells me that the establishment is already preparing for the labor displacement wave. For crypto investors, this means one thing: decentralized identity and reputation systems will become critical. If the state cannot tax AI revenue effectively, it will try to tax the users of AI — via digital identity on CBDCs. The fight over the next few years will be about who controls the on-ramp to the machine economy.
Macro is the only narrative that matters. Yang’s AI tax is a symptom, not a solution. The real question is: will the state adapt by building programmable money, or will it collapse under the weight of a tax base that no longer exists? I’ve seen this pattern before — in the 2022 Terra crash, in the 2024 ETF inflow window, in the 2025 AI-crypto convergence. The answer is always the same: liquidity flows to the most efficient, least frictioned structure. The AI tax is friction. The market will route around it.
Yang means well. But good intentions don’t alter the laws of monetary physics. The machine economy is already here. The only question is whether we architect the rails before the regulators build the walls.
