Tech companies need predictable power supplies. Energy markets are broken. Utilities are incompetent. So Meta and Microsoft are stepping in. Buying direct. Financing plants. Rescuing the sector.
The trading licence is technical minutiae. A mechanism to resell surplus when demand overshoots.
It’s theatre.
Meta locks in 6.6 gigawatts at locked rates for 20 years. Microsoft follows. The signal is unmistakable: we will consume whatever you build.
This de-risks developer capital. It keeps dying plants online. It makes the nuclear sector Meta’s responsibility. The sector stops lobbying Congress. It lobbies Meta.
But they are buying far more than they will ever consume.
Here’s the perverse incentive: Meta profits from inflating those commitments. The larger the front-loaded contract, the greater the future arbitrage margin. Forecast exponential AI demand growth. Lock in massive commitments. When efficiency gains arrive, resell the surplus at higher wholesale rates. The inflation is a feature. It maximises extraction potential.
Regulators have no mechanism to constrain this. No demand verification. No penalty for over-forecasting. When a utility overshoots demand projections, regulators investigate. When Meta overshoots on energy contracts, it’s called forward-thinking infrastructure investment. The asymmetry is complete. Meta can lie about demand with impunity. Utilities have to tell the truth. If they’re wrong, they pay. If Meta is wrong, everyone else pays.
Meta is committing public capital to infrastructure it doesn’t need based on speculative AI demand. If that demand doesn’t materialise, the public eats the cost. Meta has already extracted its margin. The plants remain. The ratepayers remain.
This is infrastructural gambling with other people’s money.
The trading licence isn’t procedural. It’s the permission to operate as a market maker.
Without it, Meta is constrained. A buyer. Contractually obligated.
With it, Meta becomes a dealer. Buy at locked rates. Sell at market rates. The gap is profit. The government legalised rent extraction from infrastructure.
In 2-3 years, chip efficiency improves. Meta’s power consumption will likely drop relative to its commitments.
Solution: resell it. The trading licence makes this legal. The margin makes it profitable.
Meta is now the largest seller. It sets prices. It forces competitors to bid at rates it decides.
What’s happening is legal. The Federal Energy Regulatory Commission granted Meta and Microsoft authority to trade power. They’re not violating rules.
That’s the problem. The rules permit this. They shouldn’t.
A trading licence was meant to hedge energy costs and manage risk. It wasn’t meant to allow corporations to front-load infrastructure financing based on speculative demand, then extract margin when that demand doesn’t materialise. Regulators have built no constraints to prevent it.
Contract rates are locked. Wholesale rates fluctuate. That spread is profit. Once you’re the largest seller, you control supply timing. Release power when prices spike. Withhold when they collapse. Extract margin from volatility you don’t create but now manage.
That’s market timing. And it’s legalised.
The structure is asymmetric. Meta controls when power enters the market. Smaller operators take whatever pricing results. Ratepayers absorb whatever rates utilities pass through. They have no trading licence. No arbitrage opportunity. No escape.
Utilities operate under public utility commission oversight. Rate-of-return regulation. Demand forecasting reviews. Stranded asset liability if they build capacity that doesn’t materialise.
Meta operates under no such framework. No demand verification. No rate regulation. No liability for stranded commitments. If Meta’s consumption projections prove wrong, the public absorbs the cost through higher electricity rates. Meta absorbs nothing.
Political pressure shifts. Nuclear plants become essential regional assets. Permitting accelerates. State capacity gets conscripted.
By 2030, tech companies will control US energy markets without regulatory constraint. Without demand verification requirements. Without liability for stranded capacity. Without obligation to prioritise grid stability over profit margin.
The fix is straightforward: treat corporate energy commitments like utility forecasts. Require demand justification. Impose stranded asset liability on over-forecasting. Restrict trading licences to actual risk hedging, not speculation.
That won’t happen. The game is mathematical. The story is moral. One obscures the other.
That’s the whole apparatus.