The sprint never stops, only the pace.
Andrew Yang just dropped a grenade on CNBC’s Power Lunch. The 2020 presidential candidate, now running Noble Mobile and co-chairing the Forward Party, isn’t talking about UBI this time. He’s talking about taxing AI instead of payroll. "Companies skip payroll taxes and healthcare costs by choosing AI over new hires," he said. The room went quiet. Then the debate started.
This isn’t a new idea. Yang built his political brand on automation warnings. In 2020, he proposed the Freedom Dividend – a universal basic income funded by a value-added tax on tech giants. He also backed crypto adoption and clearer digital asset rules. But now, the narrative has shifted. He’s targeting the AI itself, not the companies’ revenue. The logic: if a machine replaces a human, the government should tax that machine, not the human who lost the job.
Context: Why Now?
The timing is no accident. The US labor market is already feeling the squeeze. A CNBC and Generation Lab survey from August 13 polled Americans aged 18 to 34. Nearly 45% expect AI to hurt their careers. Only 10% see it as a net positive. That’s a massive confidence gap. And it’s not just fear – it’s data. Bridgewater Associates executives Greg Jensen and Nir Bar Dea wrote a New York Times op-ed estimating AI could displace 18% of current US jobs within five years. That’s roughly 30 million people. The customer service sector alone employs 2.9 million Americans, according to the Bureau of Labor Statistics. Those jobs are already vanishing.
Yang’s argument is simple: stop taxing labor. Tax the automation. He points to Anthropic CEO Dario Amodei, who floated a 3% AI revenue tax earlier this year. Amodei said the levy would apply each time a model generates revenue. Yang wants the same logic applied broadly. "It would force firms to weigh AI costs against payroll costs," he argued. The revenue would go directly to workers as checks, not retraining programs. Why? Because retraining rarely works. He pointed to coal miners and warehouse workers – past efforts that largely failed.
Core: The Technical Mechanics of an AI Tax
Let’s break this down. An AI tax isn’t a simple income tax. It’s a per-transaction levy on AI-generated output. Dario Amodei’s proposal was 3% of revenue from AI models. Yang’s version is broader – it’s a tax on the substitution of labor. The idea is to create a cost parity between humans and machines. If a company saves $50,000 by replacing a worker with an AI, the tax would eat into that saving. The government collects the difference and sends it to the displaced worker.
From my years on the exchange front lines, I’ve seen how automation reshapes liquidity. In crypto, we call it “flash crashes” when algorithms run wild. In the labor market, it’s the same thing – only slower. The tax is a friction. It’s a way to slow down the substitution rate. But the devil is in the implementation. How do you define “AI-generated revenue”? If a chatbot handles a customer service call, is that 100% AI? Or is it a hybrid human-machine system? The tax base would be incredibly hard to measure. Companies would have to report every instance where an AI performs a task that a human could have done. That’s a compliance nightmare.
Yet the data supports the urgency. The Bridgewater estimate of 18% job displacement within five years is conservative. Some studies predict 30% for white-collar roles. The customer service sector, where 2.9 million Americans work, is already seeing massive layoffs. Klarna, the Swedish fintech, replaced 700 customer service agents with an AI chatbot in 2024. The company’s CEO said the AI handled two-thirds of all inquiries. Those workers aren’t coming back.
Yang’s proposal to send tax revenue directly as checks is a smart political move. It bypasses the inefficient government retraining programs. But it also creates a new form of welfare – one tied to the pace of automation. The question is: will the tax revenue be enough? If AI displaces 18% of jobs, the tax base shrinks. The displaced workers pay less income tax. The AI tax would need to be high enough to compensate. A 3% revenue tax on AI might not cut it. Some economists have suggested a 10-20% “robot tax” to match the social cost of displacement.
Contrarian: The Unreported Angle – Crypto as the Real Solution
Here’s the angle most media missed. Yang has been a crypto advocate since 2020. He backed clearer digital asset rules and even proposed a US Blockchain Leadership Council. But his AI tax proposal ignores crypto entirely. That’s a blind spot. Why? Because decentralized AI and blockchain-based labor markets could solve the same problem without a government tax.
I’ve been tracking the AI-crypto convergence since 2025. Projects like Bittensor and Render Network are already building decentralized compute markets. They allow anyone to rent out GPU power for AI training. Smart contracts handle payments. No central authority. No payroll taxes. If a worker loses their job to an AI, they could instead contribute their own compute or data to a decentralized AI network and earn tokens. That’s a UBI built on code, not government checks.
Yang’s tax assumes the government is the only entity that can redistribute value. But crypto-native DAOs and protocols already do that. For example, a DAO could tax AI transactions at the protocol level – a 1% fee on every inference, distributed to token holders. No IRS needed. No bureaucracy. Just code. And it’s global. A US AI tax would only apply to domestic companies. But AI models are borderless. A company in Singapore could use an AI trained in India to serve US customers. The tax would be impossible to enforce.
This is where Yang’s proposal falls short. It’s a 20th-century solution to a 21st-century problem. The real innovation is in programmable money. If we want to tax AI, we should do it on-chain. Every time a model generates revenue, a smart contract takes a cut. That’s what Amodei hinted at – a 3% revenue tax – but he didn’t mention blockchain. Yang could have tied his proposal to crypto. He didn’t. That’s a missed opportunity.
From the front lines of the hype cycle.
I’ve seen this play out before. In 2020, DeFi summer was all about yield farming. Everyone thought the government would regulate it. Instead, the market corrected itself. The protocols that survived were the ones that built sustainable tokenomics. The same applies here. An AI tax won’t stop automation. It will just push it offshore. The US has already lost manufacturing to China. Now we risk losing AI innovation to Singapore, Dubai, or the EU. A tax on AI is a tax on the future.
And that’s the contrarian truth: the AI tax is politically popular but economically dubious. It punishes the very technology that could boost productivity and create new jobs. Remember, the internet killed many jobs but created millions more. AI will do the same. The real question is how to bridge the transition. Yang’s check-based UBI is a band-aid. Crypto-native solutions like data DAOs, tokenized labor, and decentralized compute are the stitches.
Speed is the only currency that matters.
Let’s zoom out. The AI tax debate is a proxy for a larger war: labor vs. capital. The traditional left wants to tax machines. The traditional right wants to deregulate. Yang sits in the middle, proposing a tax that funds individual checks. It’s a pragmatic centrist idea. But in a world where crypto moves at the speed of light, a government tax is too slow. By the time Congress passes a bill, the AI will have already replaced 10% of jobs. The market will have moved on.
I’ve been in the trenches of crypto regulation. The Hong Kong vs. Singapore battle for the Asian hub is a perfect parallel. Both cities rushed to create licensing frameworks. But the real winners were the decentralized platforms that didn’t need a license. They just launched. Same with AI. The real winners won’t be the companies that pay the AI tax. They’ll be the ones that build on-chain, tax-proof models.
Takeaway: What to Watch Next
Yang’s AI tax proposal is a signal. It means the political establishment is waking up to automation. But the crypto community should pay attention. If the government starts taxing AI, it will eventually tax crypto AI projects too. The SEC and CFTC are already circling. The next step is a “digital assets tax” on AI-generated tokens.
Here’s my forward-looking thought: watch for a crypto-native response. A DAO that implements its own AI tax, distributing tokens to displaced workers. If one emerges, it will be the real test of the concept. Not a government bill. Just code.