Andrew Yang is back. The 2020 presidential candidate, now running Noble Mobile, sat on CNBC’s Power Lunch and renewed his push for an AI tax. His argument is simple: stop taxing labor, start taxing the machines that replace it. He wants firms to weigh AI costs against payroll costs. The logic is clean. The execution is a mess — and it’s a mess that blockchain might already be solving.
Yang’s political brand was built on automation warnings. He proposed the Freedom Dividend, a universal basic income. He also backed crypto adoption and clearer digital asset rules. Now, he’s pointing to Anthropic CEO Dario Amodei’s idea of a 3% AI revenue tax. Amodei said the levy would apply each time a model generates revenue. Yang says the same logic should apply broadly. But here’s the problem: AI revenue on-chain is already being taxed by smart contracts, by gas fees, by MEV. The government is late to a party that’s already running on code.
The data is stark. A CNBC and Generation Lab survey from August 13 polled Americans aged 18 to 34. 45% expect AI to hurt their careers. Only 10% expect it to help. Bridgewater Associates executives Greg Jensen and Nir Bar Dea estimated AI could displace 18% of current US jobs within five years. They wrote a New York Times op-ed backing their own AI token tax proposal, echoing Amodei. The shift is already visible in customer service, which employs roughly 2.9 million Americans. Yang proposes sending the tax revenue directly to workers as checks. He says retraining programs rarely work — citing coal miners and warehouse staff as examples that largely failed.
Code is law, but audits are mercy. That’s the lesson from every DeFi hack I’ve analyzed since 2020. Yang’s tax proposal assumes a centralized, taxable economy. But the AI agents that will replace customer service reps are already being deployed on smart contracts. They run on Ethereum, Solana, and Base. They pay gas fees. They execute trades. They generate revenue. Taxing that revenue at the traditional corporate level is like trying to tax a tornado. The liquidity doesn’t stay in one place. It moves across chains, through bridges, into autonomous wallets.
The contrarian angle: an AI tax will accelerate the shift to decentralized autonomous organizations. If you tax AI revenue at the corporate level, companies will spin off their AI agents into DAOs. The DAO doesn’t have a payroll. It doesn’t have a CEO. It has a smart contract. Taxing that requires a blockchain-level tax — a protocol fee on every transaction generated by an AI agent. That’s technically possible. It’s also politically impossible. Governments can’t even agree on stablecoin regulation. The pool remembers what the ticker forgets: on-chain activity is permanent, but taxation is a human construct.
In my years auditing smart contracts and covering DeFi summers, I’ve seen this pattern before. Every regulation that tries to squeeze a decentralized system just pushes it further into the shadows. Yang’s proposal is well-intentioned. It’s also naive. He assumes the government can capture value from AI the same way it captures value from labor. But AI agents don’t file W-2s. They don’t have Social Security numbers. They have wallet addresses. And wallet addresses are borderless.
What the data doesn’t show: the on-chain labor shift. The 45% of young Americans who expect AI to hurt their careers are correct — but they’re also looking at the wrong careers. The jobs that will vanish are not just customer service. They’re data entry, compliance, junior auditing. Those are the same roles that crypto-native companies already automate with smart contracts. I’ve seen DAOs run entire operations with fewer than five humans. The rest is code. The tax base is shrinking, and Yang’s proposal doesn’t address that.
Speculation is just data with a heartbeat. The market is already pricing in an AI tax. Look at the tokenomics of AI-focused projects like Bittensor, Fetch.ai, or Render Network. They have built-in mechanisms for value capture — staking, burning, revenue sharing. A government-enforced AI tax would compete with those mechanisms. The result? A black market for AI labor on-chain. Companies will pay AI agents in ETH or USDC, bypassing the taxable layer entirely. The tax will only catch the compliant — the ones who still use payroll systems.
The takeaway: Yang is asking the right question but targeting the wrong infrastructure. Taxing AI revenue is like taxing smart contract execution. It’s possible, but only if you control the execution environment. Governments don’t control Ethereum. They don’t control Solana. And they certainly don’t control the AI agents that are already learning to spawn their own contracts. The future of value exchange is machine-to-machine, on-chain, and tax-agnostic. Yang’s Freedom Dividend might be better funded by a protocol-level extraction — a small fee on every on-chain AI transaction, collected by a DAO, distributed to verified humans. That’s a vision that aligns with his original crypto support. The current proposal is just a Band-Aid on a bleeding economy. The real fix is rewriting the rules before the bug writes them.
