Lake Tahoe and the Price You Can't Compete With

A utility just told 49,000 people that their electricity is being redirected to data centers. The story is bigger than anyone is reporting.

This week, Fortune broke the news that NV Energy, the Berkshire Hathaway-owned utility that has supplied most of Lake Tahoe’s electricity for decades, informed Liberty Utilities that it will stop delivering power to the region after May 2027. The reason: NV Energy needs the capacity for AI data centers being built by Google, Apple, and Microsoft in the Tahoe-Reno Industrial Center east of Reno.

Liberty serves roughly 49,000 customers on the California side of the lake. NV Energy supplies 75% of their power. Those customers now have less than a year to find a replacement.

The coverage has focused on the human story, and rightly so. Danielle Hughes, CEO of the nonprofit Tahoe Spark and a supervisor within the California Energy Commission’s Efficiency Division, put it bluntly: “We’re 49,000 customers. We have no leverage. It’s resource extraction.”

But the story missing is the economic mechanism underneath. Lake Tahoe is one of the first places in America where the Compute Heat Rate is visible to the naked eye and impacting residents.

What the Compute Heat Rate Tells Us About Tahoe

The Compute Heat Rate™ (CHR)¹ measures the maximum price of electricity at which an AI data center workload remains profitable. It is the demand-side analog of the natural gas heat rate: instead of asking how efficiently a power plant converts fuel into electricity, CHR asks how efficiently a data center converts electricity into revenue.

The blended CHR across workload types is approximately $6,350 per MWh. Frontier inference workloads (the kind running models like GPT-5, Claude, and Gemini through paid APIs) can reach $49,000 to $74,000 per MWh. Even commodity inference, the cheapest tier, tolerates electricity prices above $250 per MWh.

Now consider the residential customer in Lake Tahoe. Their price tolerance, the point at which electricity becomes unaffordable, is perhaps $150 to $200 per MWh at the extreme. Their bills have already risen 77% since late 2022, and the political backlash is fierce.

The gap between these two numbers is the entire story.

NV Energy’s spokesperson called the transition “a planned transition for many years, not a reaction to recent developments.” That framing may be technically true, but the economic analysis underlying it is interesting. What changed is not the contract; it is the value of the alternative use of those electrons. Data centers now consume 22% of Nevada’s total electricity. NV Energy’s own resource plan projects that could reach 35% by 2030. Twelve data center projects in Northern Nevada could drive 5,900 MW of new demand by 2033, according to the Desert Research Institute. That is nearly three times Northern Nevada’s current peak demand of approximately 2,100 MW.

At those volumes and at those price tolerances, 49,000 residential customers are economically invisible (but not politically, obviously).

The Demand Composition Ratchet in Real Time

What Lake Tahoe illustrates is a dynamic I have been tracking in my research that I call the Demand Composition Ratchet.

Here is how it works. When a new demand class enters an electricity market with a price tolerance orders of magnitude higher than existing customers, it does not simply raise prices for everyone equally. It changes the composition of who is competing for the same electrons. As the higher-tolerance demand class grows, lower-tolerance customers are progressively displaced: first through higher prices, then through allocation decisions exactly like the one NV Energy just made.

Each displacement makes the remaining demand pool more dominated by the higher-tolerance class. And that makes the next round of price increases or allocation decisions even more lopsided. The ratchet only turns one way as long as data centers grow in their share of the demand on the grid (which is certainly the trend).

Lake Tahoe’s rates have already risen 77% since late 2022. That was the price signal phase. What happened this week was the allocation phase: the point at which the utility stops trying to serve both customer classes on a constrained system and simply chooses the one that generates more value per electron.

What This Means Beyond Tahoe

Lake Tahoe is a small market on an isolated grid, which is precisely why the Demand Composition Ratchet became visible here. The dynamics are easier to see when you have 49,000 customers on a constrained system competing against 5,900 MW of data center demand.

But the same mechanism is operating, more slowly and less visibly, at every hub where data center penetration is rising. PJM Dominion is likely already past the threshold where CHR economics begin reshaping the wholesale market. ERCOT North is close behind. Houston is earlier but accelerating as data center operators flee the congestion in Dallas-Fort Worth.

The difference is that in those larger markets, the ratchet expresses itself through prices rather than outright allocation. Residential and industrial customers do not receive a letter from their utility telling them the power is being redirected. They receive a rate increase. Then another one. Then another one. Then, eventually, the aluminum smelter shuts down its second shift. The chemical plant moves to a cheaper grid. The paper mill closes.

Each departure makes the remaining demand pool more CHR-dominated. And the ratchet turns again.

Lake Tahoe got the letter. The question is who gets it next.

Hans Royal is the originator of the Compute Heat Rate™ (CHR) framework. All views are his own and do not represent those of any employer or affiliated organization.

1. Royal, Hans, The Compute Heat Rate: Quantifying AI-Driven Electricity Price Tolerance
and Its Implications for Wholesale Market Repricing (February 28, 2026).
Available at SSRN: http://dx.doi.org/10.2139/ssrn.6322318