Every large solar project eventually runs into the same wall, and it isn’t technology. It’s land. Panels need flat, sunny acreage, that acreage is usually already growing something or worth something, and whoever owns it wants paying either way.<\/p>\n
That argument is currently being had in British reservoir consultations and Indian cabinet meetings. The first business to settle it commercially, though, wasn’t a utility or a developer. It was a winery in Oakville, California, that worked out how much cabernet sauvignon it would have to tear out for a conventional array, didn’t like the number, and put the panels on its irrigation pond instead.<\/p>\n
Far Niente switched that array on in April 2008. Roughly 1,000 panels on pontoons over the vineyard pond, another 1,300 bolted to the ground beside it. The winery coined a word for the idea, Floatovoltaic, and still writes it with a trademark symbol on its own history page<\/a>.<\/p>\n Eighteen years later, floating solar is a global industry with gigawatt-scale plants in China and a brand-new half-billion-dollar national program in India. It started because a stone building in Napa couldn’t legally take a rooftop array.<\/p>\n The Far Niente cellar is a three-level stone gravity-flow winery built in 1885 to a Hamden McIntyre design. Napa County planning records show it was listed on the National Register of Historic Places on February 28, 1979, the same year the Nickel family began restoring it.<\/p>\n That listing removed the cheapest option. According to the case study published by the National Center for Appropriate Technology<\/a>, the building’s Register status made a rooftop array impossible without breaking regulations, and ground-mounting panels immediately around the heritage structure wasn’t an appealing alternative either.<\/p>\n So the winery did the arithmetic. Hitting its generation target meant more than 2,000 panels, and the only flat ground going spare was planted with the vines the estate is known for.<\/p>\n “Taking two acres out of cabernet production really hurts,” winemaker Greg Allen, who drove the project and retired from the group in 2024, told NCAT. That is the entire reason the pond entered the conversation.<\/p>\n The floating half is 1,000 Sharp 208-watt polysilicon panels sitting on pontoons, covering a little under an acre of water. Marine-grade cable ties them to concrete columns on the bank, which lets the whole raft ride up and down as the pond level moves through the year.<\/p>\n The other 1,300 panels are conventional ground mounts next to the water. Together they peak at about 407 kilowatts, with roughly 177 kW of that coming off the pond.<\/p>\n Thompson Technology Industries developed the floating structure and its sister company SPG Solar built the system. Neither exists in that form today: SPG was acquired by SunEdison in late 2014, according to PitchBook’s record of the deal, and SunEdison filed for Chapter 11 bankruptcy protection in April 2016. The pond outlasted both of them.<\/p>\n One detail worth getting right, because it gets mangled constantly. Far Niente was not the first floating PV system anywhere. That was a 20-kilowatt research prototype installed in Aichi, Japan, in 2007. What Far Niente built was the first commercial, grid-connected one, and that is the version everybody else copied.<\/p>\n The system needed a 500-kilowatt inverter to hit the winery’s production goal. The main service transformer feeding the property was rated for half that.<\/p>\n Far Niente’s utility requires the service transformer to be able to accept 100% of what a solar system produces, so the transformer had to be replaced before a single panel could be energized. Allen has described that as a large project in its own right, which is a diplomatic way of saying the winery had to rebuild its own grid connection before it could start saving money on electricity.<\/p>\n The build came to $4.2 million. A self-generation rebate from Pacific Gas & Electric, the 30% federal tax credit and accelerated depreciation cut the net figure substantially, and the winery covered the rest with a seven-year lease carrying a buyout option, which it exercised. NCAT reports the system paid for itself around year 14.<\/p>\n The original 500 kW central inverter lasted a decade before being swapped for 12 SolarEdge string inverters.<\/p>\n Photovoltaic cells lose efficiency as they heat up. Allen measured surface temperatures across both halves of the array and found the floating panels running up to 5\u00b0F cooler than identical units on land, which is performance nobody had to engineer or pay for.<\/p>\n Evaporation is the claim everyone repeats about floating solar, and it’s the one Far Niente has been careful with. Allen has said he believes the panels reduce evaporative loss, while also saying there is no completed study and that the winery can’t isolate the figure, because water is being pumped into and out of the pond at various times.<\/p>\n The pond is not decorative. It is the estate’s fire and frost protection system, its irrigation source and the destination for all process wastewater from the winemaking facility, with three wells topping it up. UC Davis has an ongoing research partnership with the winery on the shading and water questions.<\/p>\n The energy result is more modest than the 2008 announcements implied. Those talked about offsetting the winery’s annual usage and running a net-zero bill. NCAT’s figure is that the array covers about 80% of annual requirements, which for a 2008 install on 2007 silicon is still a solid number.<\/p>\n The resource isn’t the obstacle. NREL’s assessment of federally owned and regulated reservoirs<\/a> put the technical potential at 861 to 1,042 gigawatts DC, enough for around 1,476 terawatt-hours a year, or roughly 100 million American homes. We went through what that means for Lake Mead and the TVA system when Britain launched its own floating solar consultation<\/a>.<\/p>\n Deployment is a different picture. “In the United States, we don’t have a single project over 10 megawatts,” NREL senior legal and regulatory analyst Aaron Levine said when the study landed. NREL’s own working papers describe domestic floating PV as mostly sub-megawatt systems on closed-loop water: wastewater plants, drinking water ponds and irrigation storage ponds.<\/p>\n Most American floating solar, then, is still doing roughly what Far Niente did in 2008, at roughly the same scale.<\/p>\n The largest built system in the country is an 8.9 MW array at a water treatment plant in Millburn, New Jersey. On March 5, 2026, New Jersey’s Board of Public Utilities approved<\/a> a 10 MW project covering about 20 acres of the Wanaque Reservoir for the North Jersey District Water Supply Commission, and said it would be the largest floating solar facility in the nation. Regulators had rejected it in 2024 over Highlands environmental questions before clearing it on the second attempt.<\/p>\n Everywhere else the numbers stopped being comparable a while ago. Taiwan has a 181 MW plant that settles onto the seabed twice a day<\/a>, the Dutch have 73,000 panels on a quarry lake near Zwolle<\/a>, and on July 31 India’s Union Cabinet approved a \u20b95,070 crore program, about $530 million<\/a>, to build 5,000 MW of floating solar with 10,000 MWh of storage attached. India’s National Institute of Solar Energy puts the country’s floating potential at 102.18 GWp.<\/p>\n Back in Oakville, the hardware has become the story. The Sharp panels went in during 2007 and aren’t commercially available anymore, so individual failures can’t be replaced like for like. Allen’s estimate, given to NCAT, was around five more years of the current setup before a decision had to be made, and that clock has run out.<\/p>\n An overhaul would be smaller. Panel efficiency has moved a long way since 2007, and Far Niente has said a rebuilt system could produce the same output on roughly half the footprint, freeing up ground that would go back to vines. Which is the same calculation that put the panels on the water in the first place, just running in the opposite direction.<\/p>\n The winery hasn’t publicly announced which way it went. The pond, meanwhile, is still generating, still shading itself, and still holding the argument that every reservoir developer in India and New Jersey is now having at a thousand times the scale.<\/p>\n","protected":false},"excerpt":{"rendered":" Every large solar project eventually runs into the same wall, and it isn’t technology. It’s land. Panels need flat, sunny … <\/p>\nThe roof was the obvious answer, and it was off the table<\/h2>\n
Nearly 1,000 panels went on pontoons, 1,300 stayed on the ground<\/h2>\n
A transformer nearly stopped the whole thing<\/h2>\n
The water was doing more than holding the panels up<\/h2>\n
Eighteen years on, the U.S. still hasn’t built a big one<\/h2>\n
The pond is now the oldest problem on the property<\/h2>\n