Drones are the new carriers. Most of us are still building battleships.

Small, cheap drones have already changed war more than any exquisite weapon fielded this century. In Ukraine, commanders and officials say drones now cause roughly 70 percent of deaths and injuries on the front line, more than artillery, tanks, and rifles combined. A country with almost no navy has used drone boats to sink Russian warships and push the Black Sea Fleet back from its own coast. In the Red Sea, destroyers worth billions have spent interceptors costing about $2 million each to stop drones that may cost a few thousand dollars.

And yet across much of the defense world, drones are still treated as a novelty. A gadget. A line item. Cute. I think that's the same mistake battleship navies made about airplanes a century ago, and I think it will end the same way: the "cute" thing becomes the center of the fight, and the side that took it seriously first sets the terms.

Why I'm saying this

I've flown both sides of this argument. In the Navy I flew the F/A-18 Hornet with VFA-81, the Navy's frontline strike fighter at the time, and I was a landing signal officer, the pilot on the flight deck responsible for bringing jets aboard the carrier. That was the premier manned platform of its day.

After I left the Navy, I went to work for a defense contractor whose job was to write the training manual for the aircraft that would become the Predator. I was one of the first few pilots to fly it, before the military formally bought it. The Predator began as an Advanced Concept Technology Demonstration in January 1994, deployed unarmed over the Balkans in 1995, and by October 2001 was flying armed missions over Afghanistan. It's the aircraft that started the modern UAS era. Since then I've spent more than 5,000 hours flying manned ISR, and I have more than 7,000 hours total.

So I'm not arguing from a lab or a pitch deck. I've flown the top manned fighter, I was there at the start of the drone that changed everything, and I've spent years in the ISR world learning what it takes to put a sensor over the right spot and bring the data home. Persistent eyes over the battlefield used to need an aircraft like ours. Now a squad can carry them in a backpack.

A century ago, Billy Mitchell was in a similar spot. He was a senior airman who had commanded American air combat units in France during World War I, and he told the battleship establishment that airplanes would sink their ships. His outspokenness eventually got him court-martialed in 1925. He turned out to be early, not wrong. The history behind him is worth a few minutes.

When airplanes were cute

In November 1910, a civilian pilot named Eugene Ely rolled a Curtiss pusher off an 83-foot wooden platform on the bow of the cruiser USS Birmingham. The airplane dropped off the end, its wheels touched the water, and Ely landed on the nearest beach. On January 18, 1911, he landed aboard the armored cruiser USS Pennsylvania in San Francisco Bay. Hooks on his landing gear caught ropes weighted with sandbags: the first tailhook arrested landing. As a former LSO, I'll admit that one gets my attention.

To the people running navies, it was a sideshow. Battleships were the most expensive, powerful machines nations built, and careers, shipyards, budgets, and doctrine were organized around them. A navy's strength was counted in battleships.

The Ostfriesland

In the early 1920s, Brigadier General Billy Mitchell argued that airplanes could sink battleships. In the summer of 1921, bombing tests off the Virginia Capes put that to the test. The last target was the former German battleship Ostfriesland, a heavily compartmented ship that had survived gunfire and a mine at Jutland.

On July 21, 1921, Army bombers carrying 2,000-pound bombs went after her. Mitchell had his crews aim for near misses so the underwater blast would do the damage. The first bomb fell at 12:18 p.m. By 12:40 she had rolled over and sunk by the stern.

The Navy protested that Mitchell broke the test rules, and the official report played down bombing. The ship was anchored, unmanned, and not shooting back, so the critics had a point. But the ship was on the bottom.

Testing and training before the war

The part of this story that matters most is what came next. Parts of the Navy did the unglamorous work of finding out what aviation at sea could really do.

USS Langley, a converted collier nicknamed the "Covered Wagon," was commissioned in March 1922 as the Navy's first carrier. When Captain Joseph Mason Reeves took command of the fleet's aircraft squadrons, he found no more than eight aircraft had ever been embarked. He pushed it to 14, then supervised a loadout of 42. He wanted to know what the ship could do, not what the rules assumed.

Lexington and Saratoga joined the fleet in 1927. In the Fleet Problems, the Navy's big annual exercises, aircraft from Saratoga staged a 1929 raid that "destroyed" the Pacific side of the Panama Canal. Over the following years the carriers broke away from simply supporting the battle line, and those lessons shaped the Yorktown-class carriers authorized in 1933. None of that was combat. It was realistic testing and training against a thinking opponent, with the results written down.

When the cute thing became the capital ship

On November 11, 1940, 21 Fairey Swordfish biplanes from HMS Illustrious attacked the Italian fleet at Taranto, sinking the battleship Conte di Cavour and heavily damaging Littorio and Caio Duilio for the loss of two aircraft. On December 7, 1941, Japanese carrier aircraft wrecked the battleships at Pearl Harbor. In May 1942, Coral Sea became the first major naval battle decided entirely by air; the surface ships never sighted each other. A month later at Midway, Japan lost four fleet carriers.

From Ely to Midway was about 32 years. The airplane went from a stunt to the thing that decided who controlled the ocean.

The cost-exchange problem

Today's equivalent of the Ostfriesland isn't one dramatic sinking. It's arithmetic. Reporting from the Red Sea put the cost of a single destroyer-launched interceptor around $2 million, against one-way attack drones estimated at a few thousand to a few tens of thousands of dollars. The ships won those engagements. But a defense that spends a fortune on every cheap shot it stops is a defense someone will try to exhaust.

Over land, the same math plays out in Ukraine, where cheap FPV drones hunt armor and soldiers that cost far more to field, and small ISR drones give units persistent eyes that once required a dedicated aircraft. The high-end platforms are still powerful. They're just no longer unchallenged.

The old platforms don't vanish

Battleships didn't disappear after Pearl Harbor. They escorted carriers and bombarded beaches for decades. What changed was the center of gravity. The fleet was rebuilt around the carrier, and the battleship found a role inside it.

I expect the same here. Manned aircraft, crewed ships, and exquisite sensors aren't going away; I've spent my career in them. They'll increasingly fly alongside, against, and in support of large numbers of cheap unmanned systems. The side that sets the terms will be the one that did three things early:

Doctrine. Think hard about how the new tool changes the fight, not just how it fits into the old one.

Training. Put it in operators' hands and fly it in realistic problems until people know what it can and can't do.

Testing. Measure it honestly against realistic targets before the stakes are real. The Fleet Problems, and even the disputed 1921 bombing tests, were test events at heart.

Where the work is now

That third step is where a lot of counter-drone and sensor programs get squeezed. Bench tests and simulation are necessary, but they don't fly. A C-UAS sensor needs to see real targets moving through real air, with truth data to score against. A new payload needs real vibration, power, temperature, and geometry. Dedicated range assets are in high demand.

That's part of why I think manned platforms still matter in an unmanned era. A crewed aircraft can carry a payload under test, serve as a known airborne reference, or fly repeatable profiles that let engineers compare run to run, and the crew can adjust when the test card meets reality. It's the same kind of work that had naval aviation ready when it counted.

I don't know what the Midway of the drone era will look like. But history is pretty clear about who's ready for it: the side that stopped calling the new thing cute and started testing it.