What happened
Business Insider reports that Redwood Materials is pursuing a new energy business for AI data centers, combining second-life electric-vehicle batteries with solar panels, power electronics, software, and other on-site energy sources. CEO JB Straubel said the approach is designed to address long waits for grid connections. Redwood has already installed a 10-megawatt-hour system at a Rivian factory in Illinois and built a large battery system for a Crusoe data center, according to the report.
Business Insider reports that Redwood Materials, originally known for recycling electric-vehicle batteries, is increasingly pitching on-site energy systems to AI data centers facing prolonged waits for new electricity connections.
The proposed systems would operate behind the meter, using second-life EV batteries alongside solar panels and other on-site power sources. Redwood is also packaging the batteries with its own power electronics and software.
According to Business Insider, Redwood is already installing energy storage systems at Rivian manufacturing plants, including a 10-megawatt-hour installation at Rivian’s Illinois factory. The company has also built a large battery system for a Crusoe data center.
The report attributes the strategy to CEO JB Straubel, who wrote that the central problem for AI infrastructure is “speed to power.” The article says Redwood launched its Redwood Energy division last year.
Source details: businessinsider.com ↗
Why it matters
AI data centers need large amounts of electricity, while new grid connections can take years. If Redwood’s behind-the-meter systems work at the required scale, they could give operators another way to bring computing capacity online while also extending the useful life of retired EV batteries. The report does not independently establish the systems’ cost, reliability, emissions profile, or ability to supply a specific AI data center’s full power needs.
AI data-center construction is constrained not only by computing equipment but also by access to electricity. On-site storage and generation could reduce dependence on new grid interconnections, which Business Insider reports can take years.
The strategy could also create a second use for EV batteries after their automotive service, potentially linking battery recycling with energy infrastructure. That practical benefit remains a claim of the business model rather than an independently verified result in the source.
The report does not provide pricing, customer contracts beyond the cited Rivian and Crusoe projects, total deployed capacity, performance measurements, financing details, or evidence that the systems can replace conventional grid supply for large AI campuses.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
Which component of an AI application is the machine-learning model itself?
What to watch next
The key questions are whether Redwood can deploy these systems commercially at data-center scale, how much power and storage they can provide, and what they cost compared with grid upgrades or other on-site generation. Future evidence should also clarify which AI data-center customers are involved, the systems’ operating performance, and whether Redwood’s plans represent signed deployments or primarily a new business strategy.
Watch for named AI data-center customers, signed commercial contracts, and deployment timelines. The source describes Redwood’s opportunity and some existing projects but does not establish broad availability or a completed rollout across AI facilities.
Technical disclosures should show the systems’ power output, storage duration, uptime, battery condition, charging sources, and integration with data-center operations.
Cost and environmental comparisons will be important. The source does not quantify whether recycled-battery systems are cheaper or faster than grid interconnections, new batteries, natural-gas generation, or other alternatives.
It is also unclear how much of Redwood’s activity is an operational expansion versus a sales pitch. Independent performance data and project results would help distinguish the two.