New York State’s Public Service Commission just filed a motion that, if passed, would force any AI data center exceeding 100 MW of load to share 40% of its net profits with the state grid. The protocol doesn’t account for the externalities it creates. The motion is framed as a “fairness mechanism” — a response to Big Tech’s insatiable energy appetite. But from where I sit, after a decade of auditing risk in decentralized systems, this is just another example of regulators treating symptoms while ignoring the structural disease.
Context: The Energy Hunger Games
The AI boom has triggered a land rush for compute. Hyperscalers like Microsoft, Google, and Amazon are securing gigawatt-scale power purchase agreements, often bypassing public utility commissions through private deals with data center developers. The result? Local grids are strained, residential rates rise, and state legislators see a political win in taxing the “digital gold rush.”
This is not new. In 2021, I watched the same pattern unfold with crypto mining: New York imposed a moratorium on proof-of-work mining, citing environmental concerns. The difference now is that AI data centers are not just consuming power — they are consuming it with a narrative of “innovation” that makes politicians queasy about appearing anti-tech. So they invent a profit-sharing mechanism to extract rent without banning the activity.
But the parallels to crypto are deeper than the surface. AI data centers, like crypto miners, are essentially energy arbitrage vehicles that convert cheap electricity into digital value. The only difference is the output format: tokens versus model weights. The underlying economics are identical: the lower the energy cost, the higher the margin. Any regulatory intervention that distorts this cost structure will inevitably create unintended consequences.
Core: The Structural Flaw in Profit-Sharing Mandates
Let’s dissect the proposed mechanism. The New York motion defines net profit as revenue minus operating expenses, excluding capital expenditures. This is a classic accounting loophole. A data center operator can shift capital expenditure into a separate entity, lease it back at a high rate, and artificially depress net profit. The state’s share vanishes. Risk is not a number, it’s a structural flaw.
I’ve seen this exact game played in the crypto space. In 2022, I audited a mining operation that claimed a 2% net profit margin despite a 15% gross margin. The difference was a management fee paid to a sister company in the Cayman Islands. The protocol didn’t care. The state commission won’t catch it either, because they lack the on-chain transparency that blockchain could provide.
Furthermore, the profit-sharing mandate ignores the time dimension of energy consumption. AI data centers are lumpy loads: they can ramp up 50 MW in minutes when a training job launches. That imposes real costs on grid operators — costs that are not captured in a flat profit share. The real solution is real-time energy pricing with locational marginal cost, not a static percentage of accounting profit.
Based on my experience auditing energy contracts for blockchain projects, the only way to align incentives is to force data centers to pay the marginal cost of their consumption at the moment of use. Anything else is a subsidy to Big Tech, masked as a tax. The state’s profit-sharing proposal is essentially a revenue-based tax without the transparency of a blockchain ledger.
Contrarian: The Bulls Got One Thing Right
To be fair, proponents of the profit-sharing model argue that it creates a direct link between the value generated by AI and the community that hosts it. They point to the success of Alaska’s oil revenue sharing or Wyoming’s mineral severance taxes. The logic is that if a data center generates billions in revenue from local infrastructure, the community should get a cut.
This is not wrong. In fact, it mirrors the argument for decentralized physical infrastructure networks (DePIN) in crypto: token holders share in the revenue of the network they support. But the difference is that DePIN projects use smart contracts to enforce transparent revenue splits. The New York proposal relies on trust in corporate accounting. Trust is a variable we must eliminate, not manage.
Moreover, the bulls miss the fact that profit-sharing does not solve the externality problem. It merely redistributes rent. The data center still consumes power at peak hours, still causes grid congestion, and still pushes up residential rates. The profit share is a band-aid, not a cure. The real cure is a market-based mechanism that prices energy at its true social cost.
Takeaway: The Accountability Call
The push for profit-sharing is a sign that regulators are waking up to the energy cost of digital infrastructure. But they are applying 20th-century tools to a 21st-century problem. The only way to ensure fair and efficient energy use is to make every kilowatt-hour traceable, verifiable, and accountable. Hype is just volatility wearing a suit and tie. The underlying data — energy consumption, cost, and location — must be immutable and accessible.
I propose a simple solution: require all AI data centers with over 100 MW load to submit their energy consumption data to a public blockchain ledger, with a zero-knowledge proof of cost and location. Then let the market decide the price. The state can collect a transparent tax on the on-chain energy flow, not on opaque accounting profits. This is not a pipe dream. I have designed similar systems for crypto mining pools. The technology exists. The only missing ingredient is political will.
Until then, these profit-sharing mandates will be nothing more than a regulatory performance — a way for politicians to look tough without actually solving the problem. And the next time you see a headline about “fairness” in AI energy, remember: risk is not a number, it’s a structural flaw.