AI power race shifts leverage from chips to the grid, tightening electricity supply
The AI power race is shifting leverage away from chipmakers like NVIDIA toward utilities, grid operators, and power producers as electricity becomes the key bottleneck for data centers. CryptoSlate reports that US demand is rising faster than grid capacity, while new capacity takes years to approve and build.
In Texas, the Electric Reliability Council of Texas (ERCOT) voted on June 2 to overhaul how it admits large power users to the grid, facing a backlog that includes data centers and crypto mines. At the same time, New York lawmakers in Albany moved to pass a one-year moratorium on new large-scale data centers, potentially pausing expansion.
Goldman Sachs expects US data-center power demand to rise from 31GW (2025) to 41GW (2026) and 66GW (2027), with only about 50%–60% of scheduled capacity arriving on time due to delays and cancellations. The International Energy Agency projects data-center electricity use will roughly double by 2030, while AI-focused demand could triple, citing constraints like transformers, gas turbines, and slow grid connections.
The article highlights Texas Senate Bill 6 and ERCOT’s “pay your own way” model, which loads interconnection costs onto large customers and forces stand-down during emergencies. For traders, the key takeaway is that higher power constraints can raise operating costs for AI infrastructure and tighten margins for power-intensive compute, while also changing the price dynamics for crypto miners that rely on interruptible/cheap electricity.
Overall, the AI power race is making electricity the scarce asset that determines who scales—and who gets priced out—turning grid policy into a market-moving factor for crypto and digital-asset infrastructure.
Neutral
This news is best viewed as neutral for crypto because it is a real macro/operational constraint story, but it does not directly change crypto’s demand fundamentals overnight.
What matters for traders is the direction of second-order effects:
- Short term: ERCOT/Texas and potential New York moratorium proposals signal near-term friction for new power-intensive compute (including some mining). That can tighten miner profitability if “firm power” becomes more expensive while access rules get stricter—an effect that previously appeared in earlier power-crunch episodes when miners chased cheaper/interruptible electricity.
- Medium/long term: If grids can’t add capacity fast enough, both AI data centers and miners may face structurally higher power costs or slower expansion. That can pressure sectors tied to electricity-intensive infrastructure, but it can also increase the value of flexibility (where miners can curtail load) and incentivize on-site generation.
Historically, crypto volatility tied to miners has tended to be episodic and sentiment-driven, with prices reacting more to BTC liquidity and macro flows than to utility policy alone. Here, policy and capacity constraints are meaningful, yet the article frames them as a gradually tightening constraint rather than a single exogenous shock to BTC supply or ETF flows. Hence: neutral—watch for follow-through in miner economics and regional power pricing, but avoid assuming an immediate bullish/bearish BTC catalyst.