California’s microgrids show how localized solar, storage, and hybrid systems can protect critical infrastructure as AI-driven power demand strains the grid.

California’s experiment in microgrids offers a glimpse of how critical facilities—from tribal reservations to airports to data centers—might survive an era of surging electricity demand and increasingly fragile grids.

U.S. data center demand is projected to rise to 75.8 gigawatts in 2026, according to 451 Research, part of S&P Global. That figure represents total power demand for IT equipment, cooling systems, lighting, and auxiliary infrastructure. The scale of growth is striking. In 2025, utility power supplied to hyperscale, leased, and crypto-mining data centers totaled 61.8 GW. The jump to 75.8 GW would represent roughly 23 percent growth in a single year.

That surge follows a 22 percent increase in 2025 over 2024. For perspective, the United States had only about 25 GW of operating data center capacity in 2024, according to Bloom Energy. The primary driver of this expansion is artificial intelligence workloads, with a handful of dominant hyperscalers projected to spend aggressively on infrastructure build-outs.

As AI systems scale and computational intensity rises, so does the pressure on the electric grid. Utilities must serve not just households and industry but server farms that can rival small cities in their power draw. Against that backdrop, the microgrid is shifting from niche innovation to strategic necessity.

What Is a Microgrid?

A microgrid is a cluster of interconnected loads and distributed energy resources that functions as a single controllable entity relative to the larger grid. It can connect to the main utility network or disconnect and operate independently in what is known as “island mode.” That flexibility allows microgrids to improve reliability and resilience during grid disturbances.

Advanced microgrids combine traditional generators, renewable sources such as solar, and battery storage systems. When the larger grid fails—or in remote regions where no such grid exists—these localized systems can continue supplying electricity. Beyond resilience, they can also optimize costs, extend energy supply duration, and generate revenue through participation in energy markets.

What follows are five California-based microgrids that illustrate how these systems operate in practice and what results they deliver. Though diverse in scale and structure, they share a common goal: keep critical services running when the broader grid falters.

1. Blue Lake Rancheria Microgrid

In Humboldt County, on the tribal lands of Blue Lake Rancheria in Northern California, a modest solar installation evolved into a nationally recognized resilience model. The system began as a roughly 420 kW solar photovoltaic array paired with a battery storage system rated at about 500 kW/950 kWh. It was later expanded to approximately 1,150 kW of solar and 1,950 kWh of storage capacity.

The microgrid includes a centralized management system that serves as its controller, as well as protective relays that serve as points of common coupling. Through a computer-controlled breaker, it can island from the grid when necessary. Connected to PG&E’s 12.5 kV distribution system, it serves a campus of roughly six buildings, including government offices, a hotel and casino, and essential community infrastructure.

Its importance became clear during California’s 2019 statewide outage. While much of the region went dark for days, Blue Lake Rancheria continued to deliver electricity to water systems, sewage treatment facilities, and emergency shelter facilities. Over time, the system has produced significant reductions in greenhouse gas emissions and annual electricity savings in the hundreds of thousands of dollars.

Yet resilience is not limitless. Backup duration depends on solar generation and stored energy. Extended low-sun periods reduce independence, and capital costs remain substantial. Interconnection rules, tariff structures, and technical complexity create additional hurdles. As with most renewable-plus-storage systems, weather ultimately sets the outer boundary of performance unless additional generation is layered in.

2. Borrego Springs Microgrid

Borrego Springs, a remote town in San Diego County, was once served by a single transmission line—an obvious vulnerability in wildfire-prone terrain. Today, it’s a microgrid, owned and operated by SDG&E, that integrates two battery storage systems, a large third-party-owned solar array, two generators, and an ultracapacitor, all coordinated by advanced switching infrastructure.

The addition of local generation and storage allows critical services—fire and police stations, among them—to remain operational during transmission failures. Upgrades aimed at achieving 100 percent clean energy during islanded operation include advanced inverters and improved energy management systems that can smooth voltage swings and solar variability.

For residents, resilience is not theoretical. The town’s isolation means outages can quickly become crises. Local generation and storage help prevent a total blackout.

Still, the system illustrates the technical challenges of high-solar environments. Voltage fluctuations and output variability require sophisticated inverter design and real-time dispatch management. Storage capacity, solar production, and weather conditions impose natural limits. Backup generators remain part of the architecture to guarantee supply when renewable output dips.

3. Redwood Coast Airport Microgrid

Also located in Humboldt County, the Redwood Coast Airport Microgrid serves multiple customers, including the airport itself and the U.S. Coast Guard. This front-of-meter system pairs approximately 2.2 MW of solar photovoltaic capacity with a 2 MW/9 MWh battery storage installation built around Tesla Megapacks.

The microgrid is equipped with protection and isolation devices that allow it to island during outages. PG&E owns and operates the circuit and manages control during islanded conditions.

Notably, the system operates without fossil fuels in its regular generation mix, aligning with California Energy Commission and CARB emissions mandates. As a multi-customer, front-of-meter model, it offers a blueprint for similar deployments across the state.

But scale comes at a price. Large solar arrays, battery systems, protection schemes, and interconnection infrastructure require significant capital. Cost recovery depends on tariff design and regulatory approval. A 9 MWh battery offers meaningful coverage, yet extended outages or low-solar days still expose duration constraints. Seamless transitions between grid-connected and islanded modes demand rigorous engineering and operational discipline.

4. Calistoga Resiliency Center

In Napa County, the Calistoga Resiliency Center represents a hybrid approach that blends lithium-ion batteries with hydrogen fuel cells. With approximately 293 MWh of storage and peak output near 8.5 MW, the system is designed to deliver at least 48 hours of continuous power during Public Safety Power Shutoff events.

Its energy management platform, VaultOS, is technology-agnostic, supporting black-start capability and grid-forming functions while coordinating operations whether connected to PG&E’s grid or operating independently.

Hydrogen fuel cells extend duration beyond what batteries alone can provide, addressing one of the central limitations of renewable-storage microgrids. Yet hydrogen introduces its own complexities. Reliable supply chains, storage safety, transportation logistics, and higher upfront costs all add layers of operational responsibility. The system is robust but not easily replicated in smaller communities without significant capital and technical expertise.

5. City of San Diego Multiple Facility Microgrids

Rather than building one centralized system, the City of San Diego deployed distributed microgrids across eight municipal sites, including fire stations, police stations, and recreation centers. Together, they incorporate roughly 930 kW of solar photovoltaic capacity and approximately 2,175 MWh of battery storage.

Each facility can island during outages, with control systems enabling load shifting and dynamic switching. This distributed approach spreads risk and enhances resilience across multiple civic nodes.

The benefits extend beyond emergency preparedness. Solar-plus-storage configurations lower utility bills, manage peak demand, and integrate with electric vehicle infrastructure. Because the facilities are dispersed, the model is replicable for similar civic buildings elsewhere.

Still, distributed architecture introduces managerial complexity and may sacrifice some economies of scale. Individual storage capacity at each site remains modest during prolonged outages. Costs ultimately flow through municipal budgets or ratepayer structures, underscoring the need for clear regulatory and financial frameworks.

The Broader Lesson

California’s microgrid portfolio demonstrates that resilience is no longer optional. It is engineered, financed, regulated, and refined over time. Each system balances three variables: cost, duration, and complexity. Solar variability, storage limits, hydrogen logistics, and regulatory hurdles—none of them disappear simply because technology advances.

As AI-driven data center demand surges toward 75.8 GW and beyond, the strain on centralized grids will only intensify. Microgrids offer a decentralized counterweight: local generation, local storage, and localized control. For critical facilities—from tribal governments to airports to future AI campuses—the lesson is clear. Resilience is built in layers.

California’s experience suggests that the question is no longer whether microgrids are viable. It is how quickly they can scale before the next outage tests the system again.

This article was originally posted in EEPower.