The proposal was bold to the point of audacity. A 9.5-kilometre breakwater would impound 11.5 square kilometres of seabed. Sixteen bidirectional turbines would produce up to 320 megawatts at peak, enough to power around 155,000 homes for fourteen hours a day. The seawalls would be built to withstand a 500-year storm, doubling as coastal flood defence. The promoters even proposed using Roman concrete - the same long-lived recipe that has kept ancient harbour walls intact for two thousand years - because the lagoon was meant to outlast everything around it. In June 2015, the UK government granted development consent. Swansea, a city that had built its fortune on copper, coal, and ships, prepared to host the prototype for a new global industry.

Then came the spreadsheet. Tidal power's load factor - the share of the time it actually generates compared to its nameplate capacity - works out to about 19 percent. Offshore wind sits closer to 50. The price per megawatt-hour the project needed to be viable was high, and falling offshore wind costs made it look higher every year. On 5 June 2018, the Welsh Government offered to put £200 million of its own money on the table to save the scheme. Twenty days later, on 25 June, the UK Department for Business, Energy and Industrial Strategy rejected the necessary contract for difference. The reasoning was austere: a series of six lagoons would cost the average British household an additional £700 between 2031 and 2050, compared with the same energy from wind and nuclear. The Treasury had run the numbers and the numbers said no.

Charles Hendry, a former energy minister whose independent review had backed the technology in 2017, accused ministers of cherry-picking. "Swansea would just be the start," he wrote on the day the decision came down. "Selective quotes from my Review do not enable us to have a proper debate." The argument cut both ways. Aurora Energy Research's modelling, used by the National Infrastructure Commission, concluded that tidal power simply never becomes competitive without subsidy. Short time horizons favoured wind; long ones favoured tidal. Both sides could prove their case by choosing where to draw the line. Soon after the rejection, news emerged that the company behind the project, owned by Mark Shorrock, was £23 million in debt and had spent £37 million on a proposal that would never break ground. One of its investors was Good Energy, then led by Shorrock's wife Juliet Davenport, which held a uniquely privileged charge over the project's assets.

And yet the sea does not stop rising. In February 2019, The Guardian reported that a revived Swansea proposal was attracting interest without needing government money. Property developers, Cardiff Airport, the Berkeley Group: corporate buyers willing to sign power purchase agreements directly. By 2023, Welsh news outlet Nation.Cymru was reporting renewed momentum for a multi-billion-pound lagoon project. The original plan died on a Treasury value-for-money calculation, but the bay itself remains, doing what bays do. Twice every day the water still climbs ten metres up the sand and falls back. Somewhere in that endless vertical commute is enough energy to power a small country, waiting for an arithmetic that finally agrees.

From a few thousand feet above Swansea, the bay reads like a topographic experiment. To the south, the long curve of Mumbles Head sweeps out into the Channel. To the north, the city of Swansea climbs the slope from the docks. Between them, the proposed lagoon footprint occupies roughly the inner third of the bay, anchored at Swansea Docks on one end and curving south-east toward Port Talbot's steel works. On a low tide, you can still see why this place looked perfect: vast flats of sand, kilometres of exposed seabed, a tidal range that would humble most coastlines on Earth.