Hook
Andrew Yang just dropped a tax bomb on CNBC’s Power Lunch. His target: artificial intelligence, not payroll. The 2020 presidential candidate, now CEO of Noble Mobile, argued that the government should tax AI every time a model generates revenue—shifting the burden from human labor to machine output.
I saw the wire tap before the wallet drained. The same logic applies here: Yang is signaling a structural shift in how the state extracts value from innovation. But the blockchain community needs to understand the real vector—this isn’t about fairness. It’s about control. And the data shows the attack surface is already compromised.
Context
Yang built his political brand on automation warnings. His 2020 campaign proposed a Universal Basic Income called the Freedom Dividend, backed crypto adoption, and called for clearer digital asset rules. Now he’s reviving that narrative with a specific tax proposal: force companies to pay a levy on AI-generated revenue, effectively making the cost of automation comparable to the cost of hiring a human.
He’s not alone. Anthropic CEO Dario Amodei floated a 3% AI revenue tax in 2025. Bridgewater Associates executives Greg Jensen and Nir Bar Dea published a New York Times op-ed estimating AI could displace 18% of current US jobs within five years, and they backed an “AI token tax” as a solution. A CNBC/Generation Lab survey of 18–34 year olds found 45% expect AI to hurt their careers. The narrative is hardening.
But here’s what the mainstream misses: the tax debate is already embedded in the code. Cryptocurrency networks have been taxing economic activity through gas fees, protocol fees, and MEV (Maximal Extractable Value) for years. The question isn’t whether to tax AI—it’s who controls the tax mechanism and whether the revenue flows to decentralized stakeholders or centralized treasuries.
Core
Let’s break down the technical architecture of Yang’s proposal. He wants to tax AI “each time a model generates revenue.” That’s a transaction-based levy, similar to a per-block fee in a blockchain. But unlike a blockchain, AI models don’t have a transparent ledger. You can’t easily audit when a model generates revenue. The tax base is opaque, and enforcement would require either government API access or mandatory reporting—both of which create a central point of failure.
From my experience auditing DeFi protocols, I’ve seen how tax policies can be gamed. In 2022, I traced a phishing campaign that exploited a Telegram group to steal Ethereum user funds. The attackers used a mixer to hide the transaction trail. If the government tries to tax AI revenue, the first wave of evasion will be obvious: offshore model hosting, revenue splitting through shell entities, and tokenized revenue streams that bypass fiat rails.
Yang’s proposed solution—sending the tax revenue directly to workers as checks—is equally flawed. He correctly notes that retraining programs for coal miners and warehouse workers largely failed. But direct cash transfers don’t solve the structural problem: displaced workers need new roles, not just income. The crypto space offers a better model: DAOs that distribute governance tokens to contributors, allowing them to earn a stake in the protocol’s future success. That’s a form of universal basic income, but with upside, not a handout.
Bridgewater’s estimate of 18% job displacement is conservative. I’ve seen on-chain data from AI-powered trading bots that already execute 60% of volume on some decentralized exchanges. The customer service sector employs 2.9 million Americans—AI chatbots are already replacing junior agents. The speed of displacement is accelerating. Speed is the only currency that doesn’t depreciate.
Contrarian
Here’s the angle no one is reporting: an AI tax would actually accelerate the shift to decentralized, non-custodial AI models. If the government taxes proprietary AI like OpenAI’s or Anthropic’s, developers will migrate to open-source models running on decentralized compute networks like Akash or Render Network. These platforms don’t generate “revenue” in the traditional sense—they use tokenized payments that are hard to trace and tax.
Think of it as a governance exploit. Yang’s proposal assumes a centralized tax authority can monitor every AI revenue stream. But blockchain-based AI agents can operate pseudonymously, sending and receiving value through smart contracts. The tax base becomes invisible. I don’t trust governments to tax innovation. The crash wasn’t the crash—it’s the policy response that breaks the system.
Moreover, the 3% revenue tax is a beachhead. Once the government establishes the principle that AI-generated revenue is taxable, the rate can be raised. Look at what happened to crypto mining: the US Treasury’s 2023 proposed 30% tax on mining energy costs was a warning shot. The same logic applies to AI. The tax is leverage waiting to be wielded.
Takeaway
Yang’s AI tax push is a signal that the regulatory landscape is shifting. But for crypto natives, the real opportunity is in building AI systems that are intrinsically tax-resistant—not through evasion, but through decentralized governance. The next generation of AI agents will be DAO-controlled, with revenue distributed to token holders, not to a government treasury.
Will the next AI boom happen on-chain, off the taxman’s radar? I’m already positioning my signals accordingly. While you read the news, I traded the rumor. The question isn’t whether to tax AI—it’s who gets to define the tax base. And in a world of smart contracts, that power is up for grabs.