Why General Motors Is Building Batteries for Power Plants, Not Just Cars

General Motors has shifted its battery manufacturing plants to make large energy storage systems for electrical grids — the massive power networks that deliver electricity to cities and regions. The company is betting that artificial intelligence data centers, which consume enormous amounts of electricity, will become its biggest customers.

GM's announcement focuses on sodium-ion batteries, a type of battery chemistry that costs less to make than the lithium-ion batteries found in most electric cars. Sodium is also abundant and easier to source globally than the rare materials used in traditional car batteries.

The Problem That Created an Opportunity

AI data centers consume so much electricity that power grids in North America, Europe, and Asia are struggling to keep up. When data centers need to run heavy computing tasks — training AI models or processing millions of queries — they draw sudden, enormous amounts of power. Grid operators increasingly require these data centers to have large battery backup systems on site, the way hospitals have backup generators.

The battery storage gap is real and growing. Traditional battery makers cannot manufacture storage systems fast enough to meet demand.

Automakers like GM, meanwhile, have a different problem: they built factories to make batteries for electric cars, but fewer people are buying EVs than manufacturers expected in the early 2020s. This left expensive manufacturing plants operating below capacity. Rather than shut down or underutilize these facilities, GM is redirecting them toward grid storage — the same equipment and expertise, redeployed to a different customer.

Sodium-ion batteries turn out to be well-suited to this application. The manufacturing processes are similar to those for lithium iron phosphate, a battery chemistry that GM already knows how to produce. For grid storage, where size and weight are not concerns — unlike in a car — the fact that sodium-ion batteries are heavier matters little.

How Much Power Do Data Centers Actually Need

A single large data center — one processing millions of AI queries — can require more than a billion watts of stored energy to meet grid obligations. These obligations include maintaining grid stability, responding to power emergencies, and smoothing out peaks in electricity demand. Multiply that across the dozens of massive AI data centers being built globally right now, and the market for grid storage is enormous.

Until now, China has dominated the manufacturing of grid storage batteries, mostly lithium iron phosphate systems. The US and other governments, concerned about supply chain vulnerability in critical infrastructure, have created openings for domestic manufacturers. A grid storage system built in North America or Europe, with warranties and long-term service support, now looks attractive to data center operators evaluating their options.

GM's actual opportunity is not simply making cells. It is offering complete energy storage systems — tested, certified, backed by warranty and maintenance support — that a data center's procurement team can choose with confidence.

The Technical Tradeoff

Sodium-ion batteries currently on the market can survive about 4,000 to 6,000 complete charge-and-discharge cycles. For grid storage, where a battery system is expected to work reliably for 10 to 15 years, that performance is competitive. Lithium iron phosphate systems return slightly more electricity to the grid than they consume in charging — around 93 to 95 percent efficient. Sodium-ion systems are a bit less efficient, returning about 88 to 92 percent. At the scale of a data center, that difference in efficiency adds up to real money on the electricity bill, and engineers will need to improve it.

The engineering challenge is not new. It is precisely the kind of problem that automotive battery engineers, with decades of experience in cooling systems and power management, are equipped to solve.

A Pattern We Have Seen Before

The technology industry has navigated this kind of shift before. In the late 1990s and early 2000s, companies that made hard disk drives for personal computers — like Seagate and Western Digital — watched as PCs became cheap commodities. They did not disappear. Instead, they reinvented themselves as suppliers of massive storage systems for banks, tech companies, and enterprises. They brought manufacturing scale and deep engineering expertise, but they also had to learn to sell and service in a completely different way.

Automakers entering the grid storage business face a similar challenge. The battery engineering expertise transfers fairly cleanly. Learning to sell to utility companies and data center operators, building the service relationships and proving long-term reliability to skeptical procurement teams — that is the harder part.

What Comes Next

For data center operators and grid managers, having more domestic suppliers to choose from is straightforwardly good news. It reduces the risk of depending on a single supplier or geography. It also expands the pool of equipment that qualifies for government incentives that favor domestically made products.

The real question is timing. Grid storage systems require careful testing and official certification before they can connect to the electrical grid. Data center operators with urgent need are building fast. Whether sodium-ion batteries from automotive manufacturers can jump through all those regulatory hoops quickly enough to meet current demand — that remains to be seen.

Looking at the bigger picture, something interesting is happening. Automakers built massive battery factories to support electric vehicles. That industry is maturing more slowly than expected. But those same factories, with their expertise in power systems and thermal management, are now enabling the AI infrastructure that is reshaping computing. It is not a scenario anyone planned five years ago. But the factories are finding new purpose instead of sitting idle, and that practical redeployment of industrial capacity may turn out to be one of the significant industrial shifts of this decade.