The Long Sunrise

It begins with a fire that dies if no one feeds it. For half a million years, that was all the power we had.

A human body uses about a hundred watts — roughly a bright light bulb. That was our limit: our own muscles, then animals, then wind and falling water. For hundreds of generations, that limit barely changed.

How much energy you had shaped everything — how long you lived, what you could build, whether night belonged to you. Light was not a small comfort. Light was wealth.

Then we learned to burn old sunlight.

Coal, oil, and gas are sunlight that fell on plants long before people existed — buried underground, then burned a million times faster than it formed. In about two hundred years, energy use per person rose roughly ten times. The small fire became a furnace. The same picture appeared on three continents: smoke over the roofs, and under the smoke, people living better than anyone before them.

The cost showed up in the air. Carbon dioxide is now 432 parts per million, and still rising. Even so, rich countries began to need less energy per person. In the United States the peak was 1979; use has fallen 15–20% since, while the economy doubled. What people wanted was never the fuel itself. It was cold food, a bright page, a heavy load moved. Getting more life from less fuel is progress too.

“Too cheap to meter,” they said of nuclear power in 1954.

It did not work out that way. The more nuclear plants the world built, the more each one cost — about three times higher as more were built (Grubler, 2010). The cost of sending a kilogram into space stayed flat for forty years. Concorde was sold as the future of flight. It ended in a museum.

A closer heartbreak: Vietnam in 2020. In one year the country added about 9 GW of rooftop solar — among the fastest build-outs anywhere. Then the payment that rewarded solar power ended. The boom stopped almost overnight. Panels sat underused. Installers lost their work.

Hold onto this when the next chapters feel certain. Fast growth is not a guarantee. Things we stamp out in factories, again and again, tend to get cheaper. Huge one-of-a-kind projects, built on site under thick rules, often do not. Nothing later in this story happens by itself.

First light. From here, we stop guessing. We measure.

A solar panel cost $76 per watt in 1977. In 2025 it costs about nine cents. A battery pack cost $7,500 per kilowatt-hour in 1991. By late 2025 the pack price was $108 (BNEF) — and the cheapest packs near $50. At those prices a roof is not only shelter. It can make electricity.

In 2025 the world added 647 GW of solar. The new electricity that came with it was the largest one-year jump from any power source in history. For the first time, wind and solar together made more of the world’s electricity than coal.

Here is the hard truth — and why batteries matter. When midday solar is very cheap and very common, the grid can have more power than it needs at noon. Prices fall. Sometimes power is almost worthless for an hour. Batteries solve that. They store the cheap midday power and use it after dark. That is why large grid batteries added about 300 GWh in 2025 — up 51% in one year — from home wall units to huge battery farms on every continent.

Cheap energy runs more computers. More computers train smarter software. Smarter software helps design better machines — including machines that make energy cheaper. That loop is real. What follows marks fact and hope clearly.

What we can measure: data centers used about 485 TWh of electricity in 2025, and may reach about 950 TWh by 2030. Big technology companies spent more than $400 billion in 2025 — more than the world spent finding and drilling oil and gas. Building computers and AI now outspends the hunt for new fossil fuel. Power and chips pull each other: chips need power; power investment follows the chips.

What is still hard: clever software is not the same as clever hands. Battery and car factory lines are highly automated. Most building sites still run on people with tools. Machines learned to write and calculate before they learned to work well in dust, rain, and half-finished rooms.

We will use far more energy than we do today. That is not an accident. That is the goal. When energy is abundant, it stops being the hard limit on what we can attempt. It does not mean energy is free.

Before morning, one last look up. What follows is a dream — and we label it as one.

If machines ever build more machines in space — and none of that exists yet — solar collectors could slowly gather around the sun over centuries. Daylight becomes something people build, not only something they wait for. A full ring of collectors around the star. About twenty trillion times the power we use today.

We put no year on this. It depends on ideas we have not proven. Physics still draws a ceiling, and it is worth seeing once: