How the Lights Stay On

Chapter 7: Catching Lightning in a Bottle

Tucked into the Appalachian Mountains of western Virginia, near the West Virginia border, there is a small reservoir at the top of a mountain. There is another, larger reservoir at the bottom of the same mountain. Between them runs a tunnel, and inside the tunnel are six enormous water pumps, each as big as a house.

When electricity is cheap and abundant — at midnight, say, when nuclear plants are running steadily and nobody is awake to use the power — the pumps run. Water is lifted from the lower reservoir to the upper one. By morning, the upper reservoir is full.

When electricity is expensive — say, on a hot summer afternoon when air conditioners are humming and demand is peaking — gravity goes to work. Water is allowed to fall back down through the tunnel, spinning the pumps in reverse so that they now act as turbines. The turbines drive generators. The generators produce electricity.

This is the Bath County Pumped Storage Station, completed in 1985. It can store and release about 24,000 megawatt-hours of energy. It can produce 3,000 megawatts of power — about the same as three large nuclear reactors — for almost a full day before the upper reservoir runs dry.

It is, by some measures, the largest battery on Earth.

It is not a battery. It is a mountain.

When you hear “energy storage,” you probably picture batteries. The lithium-ion battery in your phone. The Tesla Powerwall on a neighbor’s garage. A Tesla Megapack at a utility-scale battery farm out in the desert.

Batteries are important. Batteries are growing. But for most of the last hundred years, almost all of the energy storage on the world’s electrical grids has been water in reservoirs, exactly like Bath County. Worldwide, pumped hydro still accounts for about 95 percent of all installed grid storage capacity.

The reason is simple. Pumped hydro is what engineers call a “low-tech, high-confidence” solution. The physics is gravity. The hardware is water, concrete, and turbines. We have known how to build these since the 1890s. They last fifty or eighty years with maintenance. They have no battery cells to replace, no rare metals to mine.

The problem with pumped hydro is geography. You need a mountain. You need two places to put reservoirs, one above the other, separated by enough vertical distance to store useful energy. There are only so many sites in the world that meet these requirements, and many of the best ones already have a power plant on them. You cannot just decide to build a Bath County in Kansas.

So the world has, in the last fifteen years, started building a different kind of storage. Not because batteries are better than mountains — they are not, for the things mountains can do. But because we are running out of mountains, and we need to store electricity everywhere, not just in the Appalachians.

In March 2017, the state of South Australia had a problem. After a series of blackouts caused by extreme weather and an aging power system, politicians and engineers were looking for ways to make the grid more resilient. Onto a social media argument about the situation came Elon Musk, the CEO of Tesla, who made a public bet on Twitter: Tesla would build the largest lithium-ion battery in the world for South Australia, and would do it in 100 days from the day the contract was signed — or it would be free.

The bet was real. The contract was signed. And Tesla built it in 63 days.

The Hornsdale Power Reserve, as it was named, is a battery farm in the South Australian outback. The original installation was 100 megawatts of power and 129 megawatt-hours of energy. (It has since been expanded.) Connected to a wind farm, it could store the wind farm’s output when the wind was blowing too hard and discharge it back when the wind died.

In its first year of operation, Hornsdale was estimated to have saved South Australian electricity customers about fifty million dollars by reducing the cost of “grid services” — the small, fast adjustments that keep frequency stable. It also responded to several outages on the South Australian grid faster than any conventional power plant could have. When a coal plant 1,000 kilometers away tripped offline, Hornsdale was injecting power onto the grid within 140 milliseconds. Coal plants take minutes.

Hornsdale was the moment when grid-scale lithium storage stopped being a theory and became a real tool. After 2017, batteries started being built around the world at a pace nobody had anticipated. By the end of 2024, California alone had about 13 gigawatts of battery storage on its grid — more than any country on Earth had had ten years earlier.

The reason is partly that batteries got dramatically cheaper. The cost of a lithium-ion battery has fallen by about 90 percent in the last decade. Tesla’s Megapack, the truck-sized battery that utilities now buy, costs roughly a tenth of what equivalent storage cost in 2010.

And the reason batteries are useful for grids is exactly the duck from Chapter 6. The midday belly fills the batteries with cheap solar. The evening neck draws the batteries down to meet demand. Batteries are how the grid catches the noon sun and releases it at dinnertime.

What about home batteries?

A Tesla Powerwall 3 — the home version of the same lithium-ion technology used in the Megapack — stores 13.5 kilowatt-hours of electricity. It costs somewhere between ten and fifteen thousand dollars installed, depending on your contractor. It is rated to last about fifteen years.

All 18 chapters of How the Lights Stay On · Library · All courses