The seabed here is sand. Loose, mobile sand, the kind that scours away from anything solid the moment the current finds it. RWE's engineers first tried the obvious answer: dump a 2.4-metre layer of large stones around each foundation, a stone armour ring to hold the sea back. The problem revealed itself almost immediately. The turbine piles, six metres in diameter, had to be driven straight down through that armour and deep into the seabed beneath. The stones blocked the pile. So the team switched to something almost absurdly low-tech in concept: enormous sandbags. Two layers of geotextile containers, each empty bag 1.45 by 2.38 metres, filled to 80 percent capacity on the Danish island of Rømø with a cubic metre of sand. Each bag weighed 1,400 kilograms. Stacked underwater around the foundation rings, they could be punched through by the pile driver and would then close back around the steel.

The North Sea decided to audit the design almost as soon as it was laid. In October 2013, the St. Jude storm tore across northern Europe and ran straight over the construction site. Two months later, Cyclone Xaver did the same, this time pushing a storm surge that flooded parts of the German and Dutch coasts. The seabed protection at Amrumbank West held both times. Starting in December 2013, the crews began driving the six-metre piles through the geotextile layers and into the seabed beneath, an operation more like surgery than construction - centimetre-precise on an open sea, with weather windows measured in hours.

Building the turbines turned out to be easier than getting their electricity to shore. The project ran fifteen months behind schedule, not because of the engineering on the water but because the high-voltage cable connection wasn't ready when the turbines were. Offshore wind has always had this awkward truth at its centre: a turbine that spins beautifully and produces nothing useful is a very expensive sculpture. By the time the first Siemens SWT-3.6-120 was finally installed in February 2015, the cabling delay had become its own case study in why grid infrastructure has to be planned years before the steel goes up. The full eighty turbines came online at the end of 2015, each one rated at 3.6 megawatts, each rotor 120 metres tip to tip.

On a clear day from the cockpit, the farm looks like a constellation. The turbines are not perfectly evenly spaced - the layout follows the seabed contours and the prevailing wind, with the rows angled to keep one turbine's wake from spoiling its neighbour's harvest. The North Sea here averages something close to nine metres per second of wind at hub height, which is why this stretch of water, half a day from any major port, is worth the cables, the sandbags, the cyclones, and the fifteen-month delay. From the surface, on the rare calm afternoon, you can hear nothing but the rotors whispering and the gulls. From the air, you see the geometry: eighty white columns, all turning, feeding a grid that runs back under the sea to Büsum, then south, then on to the cities that never wonder where their lights come from.