How Repurposed EV Batteries Power Data Center Growth

The batteries powering your old electric vehicle don't die — they retire. Increasingly, their second act is keeping AI running.

That's the logic behind a quietly significant project unfolding in Sparks, Nevada, where data center developer Crusoe and battery recycling company Redwood Energy are expanding a microgrid that runs on solar and repurposed EV battery packs. The expansion isn't a pilot program or a proof of concept anymore. It's a 20 MW buildout — real infrastructure, at real scale, solving a real problem that's getting harder to ignore.

What Happens to an EV Battery Before It Gets Recycled

A lithium-ion battery pack typically exits an electric vehicle when it can no longer hold about 70-80% of its original charge. For an EV driver, that degradation means range anxiety. For a stationary energy storage application — say, a grid-tied microgrid that needs to buffer solar generation and smooth out power delivery — that same battery still has years of useful life left.

The gap between "too degraded for a car" and "too degraded for anything" represents an enormous, underutilized asset class. BloombergNEF has projected that retired EV batteries will represent hundreds of gigawatt-hours of potential second-life storage capacity globally by the end of this decade. Right now, most of that material either sits in limbo or heads straight to recycling — sometimes before it's truly exhausted.

Redwood Energy's position is strategic precisely because it sits at the intersection of both outcomes. As a battery recycling campus operator, Redwood already handles end-of-life battery material. Hosting a second-life energy storage deployment on the same site closes a loop that most companies can't close — the batteries that can still perform get deployed, and the ones that can't get broken down for materials recovery. That co-location isn't incidental; it's the business model.

Why Modular Data Centers Change the Calculus

Traditional hyperscale data centers are massive, fixed capital commitments. You build the shell, pull the permits, secure the power interconnection — often a multi-year process — and then fill it with hardware. The power infrastructure has to be sized for full build-out even when you're running at a fraction of capacity on day one.

Modular data centers invert that logic. Crusoe's Spark units are self-contained, deployable in smaller increments, and can be added as demand grows. The company is expanding from four units to 24 at the Sparks site — a 500% increase in unit count — bringing total power demand to 20 MW. That kind of incremental scaling is only possible when your power infrastructure can grow alongside it.

This is where second-life batteries become more than a sustainability story — they become an operational enabler. A microgrid anchored by repurposed EV packs and solar can be sized and expanded without waiting for utility interconnection upgrades that might take three to five years in constrained markets. For a data center operator trying to deploy GPU capacity fast enough to meet AI inferencing demand, that flexibility is worth a great deal.

The servers inside Crusoe's Spark units almost certainly include NVIDIA GPUs — the hardware of choice for AI model training and inference workloads. These chips are power-hungry and in short supply. Getting them online faster, even by six months, has real economic value.

Crusoe, Redwood, and the Sparks Playbook

The Sparks project is worth examining closely because it's not a one-off experiment. Crusoe is pursuing multiple parallel strategies for power supply — including a reported 12 GWh deal with iron-air energy storage maker Form Energy — which tells you something about where this company thinks grid power reliability is headed. They're not betting on a single solution. They're building a portfolio of alternative power approaches because the conventional path (call the utility, wait for interconnection, build) is broken for anyone moving at AI infrastructure speed.

Redwood's role is equally calculated. The campus in Sparks already handles the logistics, permitting, and infrastructure for large battery systems. Hosting Crusoe's data center expansion creates a revenue stream from land and power that complements the core recycling business. It's a textbook example of industrial symbiosis — one company's operational byproduct becoming another's primary input.

The location matters, too. Nevada has been aggressive about attracting clean energy and technology investment, and Sparks sits in a region with strong solar irradiance and existing industrial infrastructure. The microgrid can generate meaningful solar production while the second-life batteries provide the storage buffer to keep compute loads running around the clock — not just when the sun is up.

Where This Goes From Here

The data center industry is facing an energy crisis it largely created. AI workloads have sent power demand projections off the charts. Goldman Sachs estimated that data center electricity consumption could increase 160% by 2030. Utilities in major markets are warning that new large load interconnections may face waits measured in years, not months.

Against that backdrop, the model Crusoe and Redwood are demonstrating has obvious appeal — but scaling it faces real constraints. Second-life battery supply is growing but not unlimited, and battery quality varies enough that stationary deployments require careful screening and testing before packs can be trusted in a critical power application. The economics work today in part because repurposed packs cost significantly less than new battery storage, but as more operators chase the same supply, that cost advantage could compress.

There's also a technical maturity question. Managing a microgrid built from heterogeneous second-life packs — batteries that may have come from different vehicle models, different chemistries, and different usage histories — is more complex than operating a homogeneous new system. The battery management software has to work harder, and the margin for error in a data center application is low.

None of that makes the approach wrong. It makes it a competitive advantage for the operators who figure it out first. Crusoe's decision to pursue this alongside iron-air storage and presumably other alternative power strategies suggests they understand that no single technology solves the problem — but a combination of approaches might.

For infrastructure investors, developers, and landowners sitting on sites with existing power infrastructure or solar potential, the Sparks project is a signal worth paying attention to. The data center market doesn't just need power — it needs power that can be deployed fast, scaled incrementally, and sourced without a three-year utility queue. Repurposed EV batteries, paired with on-site solar and modular compute infrastructure, check all three boxes. The operators who recognize that early are the ones who'll have capacity online when demand peaks — which, if current trajectories hold, is sooner than most people expect.

Call to Action: Explore how you can leverage repurposed EV batteries for your infrastructure needs at InfraSale Marketplace.

[INTERNAL LINK: EV Battery Recycling]

[INTERNAL LINK: Modular Data Centers]

[INTERNAL LINK: AI Workloads and Energy Demand]