Paraguay's $350M miner power bill implies a 96% load factor as AI gets 15-year firm power

What happened

Paraguay's state utility ANDE calls Dec. 31, 2027 the final end date for special power contracts covering miners. CoinDesk reports those contracts sell power to about 4% of global bitcoin mining at $44 a megawatt-hour, and that ANDE has just signed a 15-year deal for a Founders Fund-backed AI campus.¹

The calculation (MMOTIMES analysis)

At 944 MW running all year, miners could draw about 8.27 million MWh. At $44 that is roughly $364M, so $350M implies about 7.95 million MWh, or roughly 908 MW of average draw. That is a load factor near 96%, and an effective $42.3 per MWh on full contracted capacity. The figures are consistent with a flat tariff on a very steady load. This is our arithmetic on reported numbers, not a figure ANDE has published.¹⁴

Where the stories converge

The Scottish proposal runs the other way. A Bitcoin Policy Institute report argues Britain could use mining to absorb surplus Scottish wind and reduce payments to wind farms that are paid to stop generating. There, mining is pitched as a buyer of power that would otherwise be wasted.³

Scenario, not forecast: where firm power is scarce, long-term AI contracts may win it. Mining then survives as an interruptible buyer of surplus, and it pays for that role through a lower price or less certainty. The UK idea fits that pattern. The Paraguay figures show what is at stake in the other direction.¹³

MegaETH is a looser link. CoinDesk reports it cut 20% of staff while pursuing AI business, and that DeFi value locked on it fell from over $245M to $8.8M. That shows a crypto project pivoting, but nothing we read ties it to power markets.²

Counterarguments

The three events have no demonstrated causal link. Paraguay's decision may owe more to domestic pricing than to AI demand, and the UK paper is advocacy, not policy. Extensions could also be renegotiated, even though ANDE calls the date final.¹³⁴