Putting solar on a supermarket roof is about as controversial as putting a cart corral in the parking lot. Big flat roof, big power bill, bolt the glass down, take the credit. Retailers have run that play for fifteen years and nobody blinks at it any more.<\/p>\n
Target has been doing something stranger on 21 of its roofs. The hardware looks close enough to solar that you would walk straight past it, and it produces exactly zero watts.<\/p>\n
The panels are mirrors. Water and glycol run through them in a sealed loop, and what comes back is colder than the air outside, in the middle of the afternoon, in Texas. The heat never reaches the parking lot. It leaves the atmosphere.<\/p>\n
Target started testing them in 2021 and now has more than 600 installed across six states, with 16 more sites planned for 2026. Washington spent this summer making that a much less obvious decision than it looked a year ago.<\/p>\n
Everything on Earth radiates infrared, and most of it never gets anywhere, because water vapor and CO2 absorb it and hand it straight back. That is the greenhouse effect doing its job.<\/p>\n
But the atmosphere has a blind spot. Between roughly 8 and 13 microns it is close to transparent, so anything radiating in that band is firing into deep space, which sits at about 3 kelvin and has never once sent anything back.<\/p>\n
That is why frost forms on your windshield on nights the thermometer never hit freezing. The catch has always been daylight, when sunshine dumps roughly a kilowatt per square meter onto the same surface and swamps the effect.<\/p>\n
A Stanford group cracked it in 2014. Aaswath Raman, Shanhui Fan and three colleagues stacked seven alternating layers of hafnium oxide and silicon dioxide into a film that reflects 97 percent of incoming sunlight<\/a> while emitting hard through the 8-to-13 micron window.<\/p>\n Under direct sun above 850 watts per square meter, it settled 4.9 degrees Celsius below ambient and pulled 40.1 watts of heat per square meter. ARPA-E paid for the work, and it has been cited more than 3,300 times since.<\/p>\n Two of the authors, plus engineer Eli Goldstein, spun it out as SkyCool Systems in 2016, half an hour up the road in Mountain View.<\/p>\n The commercial version is not a photonic crystal grown in a nanofab. It is laminated plastic, US-made and PFAS-free, wrapped around a panel that behaves like a solar collector run backwards. The only electricity in the loop is a circulation pump.<\/p>\n Picking one narrow slice of the spectrum and ignoring the rest is the same logic behind those Swiss greenhouse modules that harvest only the infrared the tomatoes never use<\/a>, aimed the other way.<\/p>\n A supermarket refrigeration rack spends its life fighting one number: how easily it can dump compressor heat outside. On a 100-degree afternoon the condenser fans are shoving hot air into hotter air, and the compressors work harder for less cooling.<\/p>\n SkyCool’s panels tie into that loop through a heat exchanger and do two jobs depending on the weather. Between about 60 and 85\u00b0F they subcool the liquid refrigerant, cutting flash gas and lifting capacity. Above 85\u00b0F they switch to precooling the condenser, dropping discharge temperature and taking load off the compressors.<\/p>\n The 21 stores are in California, Florida, Indiana, Kansas, Nevada and Texas. Shawn Holzschuh, Target’s director of refrigeration, told a Food Marketing Institute conference that the CO2 store in Texas “did not go supercritical once”<\/a> the previous summer.<\/p>\n That is a bigger deal than it sounds. Transcritical CO2 racks lose efficiency badly once ambient climbs past the critical point, which is the whole reason American grocers spent years calling CO2 a northern-climate refrigerant.<\/p>\n The hard numbers from a store in Roseville, California are more useful than the anecdote. That site runs R404a, and the array added 40 kilowatts of heat rejection into the condenser circuit. Condenser capacity went up 35 percent and average discharge pressure fell 35 psi on a day that peaked at 102\u00b0F.<\/p>\n The panels lie flat, which matters on a crowded roof for the same reasons it did with that flat-mounted solar array at a Fresno water plant<\/a>. Tilt buys sun angle and costs steel, wind loading and square footage.<\/p>\n Holzschuh has said the panel system currently costs 10 to 20 percent more than simply replacing an air-cooled condenser, and that he expects the gap to close as manufacturing scales. The stated goal for 2026 is to put enough panels on one roof to delete the air-cooled condenser entirely.<\/p>\n Before Target, the California Energy Commission funded two supermarket installs and published the results. It is the least glamorous document in this story and easily the most useful, because a state agency has no reason to flatter anybody.<\/p>\n At a store in Red Bluff, an 80-panel array<\/a> added 30 to 45 kilowatts of heat rejection, dropped compressor discharge pressure by 50 psi and cut refrigeration energy use by up to 15 percent. Peak demand came down 5 to 7 kilowatts.<\/p>\n The water number gets overlooked. That roof had a sprinkler soaking the condenser coils every summer to hold capacity up. With the panels in, the sprinkler came off and the site stopped spraying tens of thousands of gallons a year onto its own equipment.<\/p>\n At the second site, in Milpitas, a 60-panel array subcooled below 75\u00b0F and precooled the rest of the year. Subcooling added 10 to 25\u00b0F and cut energy use 15 to 20 percent. Precooling added 25 to 30 kilowatts of condenser capacity. Total saved: just under 20,000 kilowatt-hours.<\/p>\n Twenty thousand kilowatt-hours is not going to reorganize the grid. It is one grocery store. It is also a measured number in a government report rather than a projection in a pitch deck.<\/p>\n Here is where the story turned this summer. On July 27, a rewritten EPA rule took effect that pushes back nearly every deadline pressuring American grocers to abandon high-warming refrigerants.<\/p>\n Under the reconsidered Technology Transitions rule<\/a>, supermarket systems now have until January 1, 2032 to hit the 150 or 300 GWP limits, five years later than the original schedule, with an interim ceiling of 1,400.<\/p>\n Cold storage warehouses got six years and an interim limit of 700. Remote condensing units got six years as well.<\/p>\n The rule also lets a supermarket raise cooling capacity by up to 15 percent without that counting as installing a new system, and EPA projects the revisions will save regulated industry more than $900 million against the 2023 version.<\/p>\n So the clock that was supposed to push American grocery chains onto CO2 and ammonia by 2027 has slipped to 2032. If you run a chain and wanted a reason to leave the old rack alone, there is now one in the Federal Register.<\/p>\n What did not change is the third pillar of the AIM Act, which caps how much HFC refrigerant can be produced or imported in the first place. That schedule still calls for a 40 percent cut through 2028, then 70 percent in 2029 and 85 percent in 2036.<\/p>\n Which is the actual argument for a mirror. The panels sit in a closed water-glycol loop and do not care what is in the pipe next to them, so the same array works on an aging R404a rack today and on a transcritical CO2 rack after the remodel.<\/p>\n Target’s engineers have made that reuse case publicly. The panels survive the refrigerant change. The condenser does not.<\/p>\n Supermarkets are the shop window. The roofs that would move the needle belong to refrigerated warehouses, which are enormous, flat, and already carrying industrial ammonia plants underneath.<\/p>\n Lineage is the largest operator on Earth and reported second-quarter results on August 5. As of June 30 it ran 498 facilities, about 87 million square feet and roughly 3.1 billion cubic feet<\/a> of temperature-controlled space.<\/p>\n It also posted its first year-over-year rise in same-warehouse physical occupancy since going public, up 90 basis points to 75.8 percent, after the industry built far more capacity than demand between 2021 and 2025.<\/p>\n Lineage has idled 15 sites, ten last year and five this year, and plans to sell around $1 billion in assets to pay down debt.<\/p>\n That is exactly the balance sheet that makes a passive, no-fan retrofit interesting. When a quarter of your racking sits empty, energy cost per occupied pallet is the number that hurts. SkyCool says Lineage moves from pilot to enterprise-scale proof in the second half of 2026, which is now.<\/p>\n Cold-store roofs have turned experimental generally. Norway went the opposite way, standing 6,400 solar panels bolt upright on an Arctic cold store<\/a> because flat glass under snow generates nothing.<\/p>\n Radiative cooling needs a dry sky. Water vapor absorbs inside the 8-to-13 micron window, so a panel in Houston in August performs nothing like a panel in Red Bluff, and cloud cover does the same thing. Nobody is putting these on a Seattle roof and calling it a chiller replacement.<\/p>\n Roof area is the other hard ceiling. A store can only reject as much heat as its square footage allows, and a 20 percent premium over a conventional condenser swap is real money across a national rollout.<\/p>\n SkyCool is also small. It says it closed 2025 at $3.33 million in revenue with $4.2 million of contracted backlog for 2026, across 30-plus installations. Those are company figures published in May through the investment platform Propel(x), not audited results, and they describe a business about the size of a regional HVAC contractor.<\/p>\nTarget has 600 of these and wants the condenser gone<\/h2>\n
California paid for the boring version, and the numbers held<\/h2>\n
Washington just made the easy option easier<\/h2>\n
Cold storage is the bigger prize, and it is having a rough year<\/h2>\n
Where the mirrors run out of road<\/h2>\n