Hook:
Andrew Yang, the 2020 presidential candidate who built his brand on a Freedom Dividend and a robot apocalypse warning, is back on CNBC’s Power Lunch. This time he’s not talking about UBI checks. He’s pushing an AI tax—a levy on the revenue generated by artificial intelligence instead of payroll. He says firms skip payroll taxes and healthcare costs by choosing AI over new hires. The logic is simple: tax the machine, not the human. But here’s what Yang misses: the tax code is already being written in Solidity, not in Washington. The real innovation isn’t a federal tax—it’s a programmable, on-chain fee that distributes value directly to the displaced. We mined liquidity while the code slept. Now we need to mine the tax.
Context:
Yang’s revival of the AI tax debate comes at a specific moment. The CNBC and Generation Lab survey from August 13 found that 45% of Americans aged 18 to 34 expect AI to hurt their careers. Only 10% expect it to help. Bridgewater Associates executives Greg Jensen and Nir Bar Dea wrote a New York Times opinion piece estimating that AI could displace 18% of current US jobs within five years. They backed their own AI token tax proposal, echoing the earlier call from Anthropic CEO Dario Amodei, who floated a 3% revenue tax on AI models in 2025. Amodei argued the levy would apply each time a model generates revenue—a per-transaction fee on machine output.
Yang’s framing is fundamentally political. He co-founded the Forward Party, runs Noble Mobile as CEO, and has always leaned into automation anxiety as a wedge issue. During his 2020 campaign, he proposed the Freedom Dividend ($1,000/month for every adult) and backed clearer cryptocurrency rules. He saw digital assets as a natural distribution channel for UBI. Now he’s pushing the tax side: send the revenue directly to workers as checks, because retraining programs (coal miners, warehouse staff) have largely failed. On the surface, it’s a clean narrative. But the crypto world has been experimenting with this exact mechanism for years—just without the government’s permission.
Core:
Let’s break down the technical structure of an AI tax, then compare it to what blockchain already does better.
A revenue-based AI tax like Amodei’s 3% proposal requires a point of collection. Traditionally, that means the government audits company financials and imposes a levy. But AI models are not like factories. They are decentralized, often running on distributed infrastructure, generating revenue through API calls, subscriptions, or embedded services. A single model might be hosted on AWS, accessed via a third-party app, and monetized through a token. The tax base is slippery. The IRS has no direct line into an OpenAI inference call.
Blockchain solves this with programmable money. Imagine a smart contract that sits between the AI model and the payment gateway. Every time a model generates revenue, the contract automatically splits the fee: 3% goes to a tax pool, 97% goes to the operator. The pool is a transparent, auditable vault. The tax can be distributed via a second smart contract that issues UBI tokens to verified human wallets. The whole system runs on-chain, immutable, trustless.
This is not science fiction. I’ve been building these kinds of automated fee structures since 2020, when I deployed $50,000 into Uniswap V2 liquidity mining. The yield distribution logic—share of fees proportional to liquidity provided—is exactly the same calculus as a tax pool. The difference is governance. In DeFi, the fee percentage is set by a DAO vote. In a government tax, it’s set by legislation. But the implementation layer is identical.
During my 2024 Spot ETF arbitrage strategy, I wrote a Python script that monitored on-chain transfers vs. exchange inflows to capture a persistent 0.5% premium. That script was essentially a tax-collection bot—it took a fraction of every trade and routed it to my wallet. The same principle applies to an AI tax. You need a reliable oracle that reports revenue events, a smart contract that executes the split, and a distribution mechanism that doesn’t rely on human intermediaries.
Yang’s proposal is backward-looking. He envisions the government taxing AI companies and mailing checks. That’s the payroll tax model—slow, bureaucratic, leaky. The crypto-native version is a real-time, permissionless system. Every time you query an AI model, you implicitly pay a micro-tax that goes directly to a network of verified humans. No forms, no audits, no delays.
This is where the data gets interesting. The Bridgewater estimate of 18% job displacement is conservative. In customer service alone, the sector employs 2.9 million Americans. AI chatbots have already replaced front-line support agents. The Bureau of Labor Statistics shows a steady decline in these roles. But the displaced workers aren’t getting retrained—they’re getting gig-economy scraps. A programmable tax could give them a direct, ongoing stake in the productivity gains of the machines that replaced them.
Let’s look at a concrete example. Suppose a company deploys a customer service AI that handles 10,000 calls per day. Each call costs $0.50 to process. The AI saves the company $5,000 per day vs. human agents. Under Yang’s proposal, the government would tax the company’s revenue and send a check to the displaced workers. Under an on-chain model, every call triggers a smart contract that deducts, say, $0.05 from the company’s revenue and sends it to a pool of tokenized human IDs. The worker gets a daily transfer—not a monthly check. The pool is transparent. The tax rate is adjustable by a decentralized governance body, not a committee in Washington.
I’ve seen this work in practice. In 2026, I launched “The Oracle’s Hand,” a copy-trading platform where AI agents execute trades based on my verified historical signals. The platform had $5 million in TVL and 2,000 active users. During a flash crash, the AI failed to pause trading, but my manual override saved 15% of the community’s funds. That experience taught me that human oversight is essential—but so is automated, transparent fee collection. The platform charged a 2% performance fee that was automatically distributed to stakers. The code never slept, but we had a circuit breaker. The same architecture can tax AI revenue and distribute UBI.
Now, the objection: “But the government needs to enforce compliance.” The beauty of smart contracts is that they are self-enforcing. If the AI model’s revenue flows through a blockchain, the tax is deducted at the protocol level. The company can’t evade it. The IRS doesn’t need to audit—the ledger is public. This is the same logic that made stablecoins resistant to censorship. Once value is on-chain, it obeys code, not borders.
Contrarian:
Let’s pump the brakes. The contrarian view is that an AI tax—whether government or blockchain—could stifle innovation. The AI industry is still nascent. Imposing a 3% revenue tax could slow adoption, push companies to obscure revenue, or drive them to jurisdictions without such taxes. The same argument was made against crypto taxation. But the reality is that taxation is inevitable. The question is who designs the system.
Yang’s approach is classic paternalism: tax the bad thing (AI replacing jobs), fund the good thing (UBI). But history shows that retraining programs fail because they don’t address the structural mismatch. Coal miners don’t become coders. The AI tax could become a permanent subsidy that keeps displaced workers in a dependency loop, rather than incentivizing new forms of human-AI collaboration.
Worse, an AI tax could be captured by incumbents. Large companies like Google and Microsoft have the resources to lobby for exemptions or loopholes. Small AI startups would bear the brunt. The same dynamic plays out in crypto: DeFi protocols with deep liquidity can absorb fees, while newcomers struggle. The tax becomes a regressive barrier.
From a blockchain perspective, the technical risk is oracle manipulation. If the revenue data comes from a centralized source (e.g., a company’s API logs), it can be faked. Decentralized oracles like Chainlink mitigate this, but they add complexity. And distribution is tricky: how do you verify that a recipient is a unique human, not a bot? Soulbound Tokens (SBTs) were proposed three years ago as a solution, but they’ve languished because no one wants their credit record permanently on-chain. An AI tax UBI would require a privacy-preserving identity system. We’re not there yet.
During the 2022 Terra-Luna collapse, I saw what happens when algorithmic trust breaks. UST was supposed to be a stable, decentralized currency. It wasn’t. The same fragility could infect an AI tax system. If the oracle fails, the tax pool freezes, and UBI payments stop. The human cost is immediate.
But the contrarian punchline is that Yang’s government tax is even more fragile. It depends on political will, which cycles every four years. A blockchain-based tax is persistent. Once deployed, it runs until the contract is updated. The code is law—for better or worse.
Takeaway:
Liquidity is just trust, digitized and leveraged. The AI tax debate is really a debate about how we distribute trust. Yang wants to centralize it in the Treasury. The crypto-native path is to decentralize it in smart contracts. We’ve already seen the model work—DeFi generates billions in fees, automatically distributed to liquidity providers. The same architecture can tax AI and fund UBI. The question is no longer whether we should tax AI, but who controls the code that does the taxing.
We rode the wave until it broke our boards. Now we need to build a new board—one that taxes the wave, not the surfer. The next narrative isn’t AI replacing jobs; it’s AI paying for the jobs it replaces. And that narrative will be written in Solidity, not in Washington.
We traded hope for efficiency, then lost both. Let’s not trade trust for convenience.