Why GM Is Building Giant Batteries for Data Centers—Not Just Cars

General Motors is redirecting its battery manufacturing plants toward a new market: grid-scale energy storage systems that power AI data centers. The company is focusing on sodium-ion chemistry, a type of battery that costs less to make and performs better in large, stationary installations than in cars—a move that signals how legacy automakers are now positioning themselves as suppliers to utilities and major tech companies running AI infrastructure.

GM's announcement centers on sodium-ion chemistry, a technology that trades some energy density for lower raw material costs, more geographically diverse sourcing of minerals, and thermal characteristics better suited to stationary storage than to mobile drivetrains.

The Supply Crunch Driving This Move

AI and machine learning workloads are consuming vast amounts of electricity. Data centers running training and inference—the computational work behind AI systems—are straining the electrical grid across North America, Europe, and Asia. Hyperscalers (the big tech companies building these data centers) and storage operators increasingly cannot secure enough battery backup and storage through traditional utility channels. The shortage is measured in gigawatt-hours, and it is growing faster than conventional battery makers can keep up.

At the same time, automakers face an awkward problem: they built massive battery manufacturing plants during the shift to electric vehicles in the early 2020s, but EV adoption has slowed in several markets. Factories that were expected to run at full capacity now sit partially idle. The equipment, manufacturing expertise, and software these plants contain—thermal management systems, battery management software, cell production lines—transfers relatively well to grid storage with modest retooling, especially if the battery chemistry is adjusted to prioritize durability and cost per kilowatt-hour rather than fitting as much energy as possible into a small, lightweight package.

Sodium-ion fits this transition well. The manufacturing processes for sodium-ion cathodes and anodes share significant similarities with those for lithium iron phosphate batteries, which most major automakers already produce or have access to. Sodium is abundant and not concentrated in a few countries the way lithium is, reducing supply chain risk. For grid applications, where size and weight are not constraints, the fact that sodium-ion cells store less energy per unit volume becomes irrelevant.

The Scale of Data Center Demand

To understand why this matters, consider the electrical demands of a modern AI data center campus. A large facility running 500 megawatts of computing power (a typical "hyperscale" installation) requires more than a gigawatt-hour of on-site battery storage. This storage handles frequency regulation (keeping the grid stable), backup power for outages, and peak-shaving (reducing demand at high-cost times). With dozens of such campuses announced or under construction globally since 2024, the total market opportunity for grid storage tied to AI is substantial.

Traditionally, lithium iron phosphate batteries—mostly manufactured in China—have dominated grid storage. But geopolitical tension and export controls on both sides of the US-China technology divide have created openings for domestic alternatives. Sodium-ion systems manufactured in North America or Europe are now competitive contenders.

This is the opportunity GM and potentially other automakers are pursuing: not just making battery cells, but selling complete energy storage systems with warranty, maintenance infrastructure, and engineering support—the full package that a major data center operator needs to evaluate against other options.

How Sodium-Ion Stacks Up

Sodium-ion cells in commercial use today typically use cathodes made of layered oxides or Prussian blue analogues (a crystalline compound) paired with hard carbon anodes. Modern sodium-ion cells can now achieve 4,000 to 6,000 full charge cycles—a lifespan competitive with lithium iron phosphate systems for utility applications designed to last 10 to 15 years.

One trade-off: sodium-ion systems are slightly less efficient at converting electricity in and out of storage. Lithium iron phosphate systems return 93–95% of the energy you put in; sodium-ion currently returns 88–92%. For a data center operator paying market electricity rates, that efficiency gap matters across thousands of charge-discharge cycles. Engineers will need to improve it, and this is exactly the kind of thermal management and battery software challenge that a major automaker's battery division is equipped to solve.

There is also a logistical advantage that rarely makes headlines but matters operationally: sodium-ion cells can be shipped in a fully discharged state without the same regulatory restrictions as lithium cells. For large grid deployments, that simplifies shipping and handling.

A Transition This Industry Has Seen Before

The broader context here is instructive. The technology industry has navigated similar transitions. Disk drive manufacturers in the late 1990s and early 2000s—companies like Seagate and Western Digital—watched the consumer PC market commoditize and shifted their focus to enterprise storage arrays and data center infrastructure. The product changed, but more significantly, so did the sales approach, service model, and customer relationships. Automakers now entering grid storage face an analogous challenge: the engineering capability transfers fairly cleanly, but building relationships with utilities and hyperscaler procurement teams is an unfamiliar business.

GM's entry into sodium-ion grid storage is, at minimum, a statement that the company views its battery business as something more than a captive supplier for its own electric vehicles. Whether that repositioning generates sustained revenue will depend on how well GM executes in a marketplace—utility and hyperscaler procurement—where automotive brand recognition carries limited weight.

What This Means for Competition and Procurement

For existing grid storage specialists and battery suppliers, the arrival of manufacturing capacity at automotive scale is a competitive factor. GM brings production scale, logistics networks, and the potential for vertically integrated system offerings that smaller storage specialists cannot easily replicate at competitive cost.

For data center developers and their power procurement teams, more domestic suppliers qualified to build grid storage systems is straightforwardly useful. It reduces dependence on a single supplier and expands the pool of eligible equipment under domestic content incentives like the US Inflation Reduction Act.

The realistic question is timing. Grid storage projects require long procurement cycles, rigorous certifications for grid-connected equipment, and testing to prove long-term reliability. Data center operators with the most urgent need are moving quickly. Whether sodium-ion systems from automotive manufacturers can achieve the necessary certifications and field-proven cycle-life data that investors and lenders require—all within a timeframe that matches current construction schedules—remains an open question.

The convergence of battery manufacturing infrastructure built for electric vehicles and the growing appetite for grid storage from AI data centers is a structurally useful accident. Capacity that might otherwise have been idle is finding a second and consequential application. That is not an outcome anyone planned from the start, but it may end up being one of the more significant industrial pivots of this decade.