A carbon dioxide battery in the Gobi takes in 290 million kilowatt hours of surplus power a year and gives back 180, which is the one ratio its own announcements never print
A steel vessel stands in a long row of identical vessels on desert gravel, frost creeping along the pipe that leaves it.
Inside is carbon dioxide, held as a liquid under pressure at roughly room temperature.
Nothing is burning on the site. Nothing is being pumped out of the ground or put back into it.
Divide the second by the first and the answer is about 62 percent, and the agency report carries both figures in the same passage without ever placing them together.
The plant sent power into the regional grid for the first time on the thirtieth of September, 2026.
Two numbers describe its working year, and both are published.
Which leaves the question nobody prints. What happens between them?
Why 1 round trip costs a third
Because every step in the loop has to hand its heat back.
Compressing a gas heats it, and that heat is not a nuisance to be vented. It is a large share of the electricity that went in, so the plant parks it in solid thermal stores until the discharge.
Then the cooled gas has to become a liquid, which means rejecting more heat to the surroundings at desert temperature.
On the way back the liquid is warmed with the stored heat, expands, and drives a turbine. Every exchanger between those steps works across a temperature difference, and every temperature difference is a loss.
So the gap between what goes in and what comes out is not a fault in the machine. It is the machine’s specification, and it is the number to ask.
5 gas holders over 139 acres
The hardware sits on about 139 acres of flat gravel in Mulei Kazakh Autonomous County.
There are 5 gas holders, 141 liquid storage units and 97 thermal storage units, and the gas side carries roughly 1.5 million cubic meters of carbon dioxide in holders that change volume as the plant breathes.
That explains why the site has no cavern. Compressed air plants at this scale usually need a salt dome or a worked out mine for the low pressure side, and this one holds it above ground.
The switch between gas and liquid happens at around 1,015 pounds per square inch and about 82 degrees, which is why the liquid needs no refrigeration plant standing next to it.
Even the start date is not quite settled. The state news agency puts grid connection on September 30, the state broadcaster puts it on October 2, and nobody has reconciled the two.
100 megawatts for 6 hours or for 10
Now the rating, and the disagreement inside it.
The plant is quoted at 100 megawatts and 1,000 megawatt hours. Divide one by the other and you get ten hours, and an account of the university team behind the technology says exactly that.
The broadcaster and the contractor that built the gas holders both say more than six hours. Those two statements cannot describe the same discharge, and the difference is whether the thousand is energy stored or energy returned.
The durability figures are less contested. More than 20,000 cycles, a design life above 30 years, and an operating range from 40 below zero to 158 degrees.
An independent write up puts cycle efficiency near 70 percent and system round trip at 66, which is the only published estimate of its kind.
What 290 against 180 works out to
Here is the arithmetic the announcements leave to the reader.
The plant is stated to absorb 290 million kilowatt hours a year and to return 180, in the way that an Austin pilot or a lab record turns into a solved problem.
Divide the second by the first and the answer is about 62 percent, and the agency report carries both figures in the same passage without ever placing them together.
The comparison abroad is smaller and claims more. An Italian developer runs a unit rated at 20 megawatts and 200 megawatt hours and advertises 75 percent over a ten hour discharge, and an independent analysis sets the two designs side by side.
The Italian firm says its design is patented in several jurisdictions and that it is taking appropriate action to defend its rights. No filing has been made public anywhere, and the Chinese supplier holds its own patents.
Where the missing 110 million goes
Strip the dispute away and the useful point is about what long storage is actually for.
A round trip in the low sixties would be unacceptable in a lithium battery, which gives back something close to nine tenths of what it takes. It is ordinary for anything that stores energy as heat and pressure.
What the loss buys is duration and materials. Steel vessels, a common industrial gas and a bed of hot solids do not care how long they sit full, and they use no lithium, cobalt or nickel.
So the honest comparison is not against a battery pack. It is against the other ways of holding a desert’s midday surplus until evening, and most of those need a mountain or a river.
A carbon dioxide loop trades efficiency for hours and for cheap parts.
Somebody still has to count the losses.
Hugo is an engineer with strong technical expertise. Multilingual from an early age, his writing combines technical clarity with a strong interest in science and energy.