Solar Outgenerated Coal in May — And the Data Points Further

Solar produced more electricity than coal in the United States for the first time in a single month, generating 47,147 gigawatt-hours in May against coal's 45,119, according to new figures from the U.S. Energy Information Administration. Solar also beat wind for the month.

The crossover was visible a decade out. Over the first five months of 2026, utility-scale solar output rose 21.6% year over year while coal generation fell 10.9%, EIA data show. The agency projects solar, wind and battery storage will add roughly 83 gigawatts of new capacity by May 2027. Fossil and nuclear capacity will shrink by nearly 4.7 gigawatts over the same period.

The economics behind the shift

The driver is a cost curve that has been bending for more than a decade. Each time the world's installed solar capacity has doubled, the cost of producing it has fallen by about 20%, according to Our World in Data. Engineers call this regularity Swanson's Law, and it descends from a rule Theodore Wright identified in aircraft manufacturing in 1936: airplane costs dropped a fixed percentage every time production doubled. The same math now applies to batteries, chips and the panels going up across Texas.

The price numbers tell the story. Utility-scale solar cost around $359 per megawatt-hour in 2009. By 2026, according to Lazard data cited by Heatmap News, it sat near $69 — an 80% decline.

Big buyers are accelerating the curve

The companies with the most to lose from constrained energy supply are moving quickly and pulling the curve forward. In February 2026, Google signed two 15-year contracts with TotalEnergies for a gigawatt of new Texas solar capacity, enough to deliver 28 terawatt-hours of electricity to its data centers. It was the largest renewable power deal TotalEnergies had ever signed in the United States.

Meta placed a similar bet a year earlier. It contracted for the entire output of Enbridge's 600-megawatt Clear Fork project near San Antonio — a $900 million solar farm built to power its data centers.

Both companies moved for the same reason, and it is not branding. Solar is now the cheapest and fastest new power they can buy, and they need staggering amounts of it for AI workloads.

A second curve stacks on top

Artificial intelligence is pushing electricity demand up faster than the grid has seen in decades. Wood Mackenzie expects U.S. solar generation to grow 65% between 2026 and 2030, with 160 gigawatts of large new power requests already in the pipeline. When a falling cost curve meets a rising demand curve, adoption accelerates.

That compounding effect matters for the next decade of power contracts. Every terawatt-hour of AI-driven demand strengthens the economics of the technology already on the steepest cost decline, which suggests the gap between solar and coal will widen rather than close.

What it means for other industries

Soren Kaplan, writing on Inc.com, argues the solar story is really a lesson in how disruption arrives — not as a sudden event but as the logical trajectory of a cost curve bending for years. The question for any leader watching a slow-moving trend is how to distinguish a curve that will reshape an industry from one that never will.

Kaplan's framework has three parts. Track slopes over snapshots: a technology's price today matters less than how fast it is falling year over year. Commit during the middle: the cheapest moment to back a curve comes before the crossover becomes obvious to competitors. Watch for stacking demand: a cost curve turns unstoppable when a second force like AI or new regulation multiplies it.

For energy markets specifically, the EIA's projections imply the next 12 months will extend May's milestone from a first into a pattern. The technologies that remake industries announce themselves years in advance, in the slope of a curve, for anyone disciplined enough to chart it.