When winter storms barrel through the Sierra Nevada, the 49,000 households served by CalPeco, the small utility that keeps the lights on around Lake Tahoe, depend on a grid that was never built for what is coming. The utility’s transmission lines and substations were sized decades ago for a mountain resort community with predictable seasonal peaks: heavy heating loads during blizzards, moderate cooling demand during tourist season, and quiet shoulder months in between. Now, AI data centers are pushing into the region, and their appetite for electricity threatens to swallow capacity that residential customers need to survive the basin’s most dangerous weather.
A grid designed for ski lodges, not server farms
CalPeco serves communities scattered across eastern California and western Nevada in one of the most weather-exposed utility territories in the West. A 2014 capacity planning filing with the California Public Utilities Commission (CPUC) described a system built around roughly 50,000 customers, with infrastructure calibrated to the residential and small-business loads typical of a high-elevation resort area. The filing laid out load-distribution strategies and approved growth projections, but none of them anticipated the kind of demand that hyperscale computing would eventually bring.
Data centers operate on a fundamentally different consumption pattern than homes. A residential neighborhood draws power in waves, spiking during cold mornings and hot afternoons, then dropping overnight. A data center draws power constantly, 24 hours a day, 365 days a year. According to the U.S. Department of Energy, a single mid-size data center can consume as much electricity as 30,000 to 50,000 homes. On a grid as small and isolated as CalPeco’s, adding even one such facility changes the math for everyone else connected to the same lines.
Every megawatt committed to a new commercial connection on a capacity-constrained system is a megawatt unavailable for residential headroom. That means tighter operating margins during peak demand, less room for error when equipment fails, and a higher probability that extreme weather will push the grid past its safe limits. For Lake Tahoe residents, that is not an abstraction. Losing power during a Sierra blizzard, when temperatures can plunge well below zero and roads become impassable, is a direct threat to life.
Climate exposure raises the stakes
The Desert Research Institute (DRI), a Nevada-based environmental science organization, has spent decades tracking temperature extremes and storm intensity across the Great Basin and Sierra Nevada. Its research documents a clear trend: the Tahoe Basin is experiencing more intense winter storms and increasingly frequent summer heat events, both of which drive sharp spikes in electricity demand.
CalPeco’s grid has historically managed those spikes because they are temporary. Equipment heats up during a peak, then cools during off-peak hours. Maintenance crews use shoulder seasons to repair and upgrade components before the next stress cycle. Layering a constant, industrial-scale data center load on top of those seasonal peaks eliminates the recovery window. The grid runs hotter, longer, with less margin for the kind of weather that the Tahoe Basin reliably delivers.
Unlike urban areas served by multiple utilities and crisscrossed by redundant transmission paths, CalPeco’s territory has few backup options. If a substation trips or a feeder line goes down during a storm, there is no neighboring utility ready to pick up the slack. That isolation is what makes the capacity question so consequential: even a modest reallocation away from residential service can have outsized effects in a system with no safety net.
The financial pressure to prioritize data centers
Utilities earn revenue through interconnection fees, demand charges, and long-term power purchase agreements. A single data center contract can generate more income per megawatt than hundreds of residential accounts combined. That financial reality creates a structural incentive to prioritize large commercial connections, even when doing so compresses the reliability margin for existing residential customers.
CalPeco’s 2014 CPUC filing gave the utility considerable discretion to allocate transmission capacity among competing uses, provided it stayed within broadly defined regulatory parameters. But those parameters were set before AI training clusters and cloud-computing campuses became major power consumers in the western United States. The planning assumptions that governed CalPeco’s grid design are now more than a decade old, and no publicly visible update to the utility’s load-allocation framework has appeared in CPUC docket records as of early 2026.
That regulatory lag is not unique to CalPeco. Utilities across the country are grappling with data center demand that has outpaced the frameworks regulators built to manage grid growth. But the consequences of that lag are sharper in a small, isolated system where the margin between adequate service and rolling blackouts is already thin.
What remains unconfirmed
Several critical details have not yet surfaced in public records. No current CPUC filing or load-forecast document quantifies the exact number of megawatts being redirected from residential circuits to data center interconnections. The figure of 75 percent, which has circulated in regional reporting, has not been traced to a specific load-flow study, engineering report, or regulatory testimony in the documents reviewed for this article. Until a primary source document emerges to support it, that number should be understood as a reported estimate rather than a confirmed engineering calculation.
The specific data center operators seeking connections in or near CalPeco’s service territory have not been publicly identified in available regulatory filings. That gap matters because the distance between a facility and the nearest substation determines how much transmission capacity it absorbs and how directly it competes with residential feeders. A data center tied into a substation that already serves multiple neighborhoods poses a very different risk than one connected to a dedicated industrial spur built with new infrastructure.
CalPeco has not released outage-modeling reports or winter-peak simulation results showing how residential service performs when high seasonal demand coincides with full data center load. Those simulations are standard practice for utilities facing major load additions, but none have appeared in publicly accessible CPUC records. Without them, it is difficult to assess whether the utility has planned adequate backup generation, feeder upgrades, or substation reinforcements to offset the capacity shift.
The contractual terms between CalPeco and its commercial customers are also opaque. Key provisions, such as whether the utility retains the right to curtail data center consumption during emergencies to protect residential service, are typically embedded in confidential agreements. Without access to those terms, the public cannot evaluate whether residential customers have any contractual protection during grid emergencies.
What Lake Tahoe residents are watching for
Local officials have voiced concern that the reallocation trades long-term residential grid stability for short-term commercial revenue, but no formal municipal resolution or legal challenge has been documented in the public record. Whether affected communities will pursue regulatory intervention through the CPUC, seek legislative action in Sacramento, or negotiate directly with CalPeco remains an open question.
The strongest evidence available today establishes a clear structural risk: CalPeco operates a constrained, weather-exposed grid originally designed for a modest resort community; AI and other data center loads represent a fundamentally new category of demand on that system; and the regulatory framework guiding capacity decisions has not been publicly updated in over a decade. The precise scale of any diversion from residential service, and whether it will translate into real-world outages, hinges on filings, modeling, and contract details that have not yet entered the public record.
For the 49,000 households that depend on CalPeco to keep the heat running during January storms and the air conditioning working during August heat waves, the question is not theoretical. It is whether the grid they pay for every month will still be there when they need it most, or whether it will be busy powering someone else’s servers.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
