Why China’s Supercritical CO₂ Turbine Matters

For over 200 years, we’ve been generating electricity essentially the same way: by boiling water. And it’s worked! Today, steam turbines power the vast majority of the world’s electricity. But this technology, which has powered us since the industrial revolution, might not be up to supplying us enough energy in this new revolutionary era of artificial intelligence. AI data centers are driving energy demand through the roof. The US alone needs to increase power output by 165% before the end of this decade just to keep up. ¹ We need to find ways to generate more power right now, without taking years to build entirely new facilities. And that’s where things get interesting.

In December 2025, China flipped the switch on the world’s first commercial supercritical CO2 power generator. It’s called Chaotan One, and it replaced the steam generator at a steel plant in Guizhou Province. The claims? An 85% increase in efficiency and 50% more power output. Same facility, better generator. ² And the full-size turbine? It’s about the size of a desk. ³

Now, that’s a series of bold claims. And I’m genuinely excited about the potential here. But this is not my first emerging energy technology rodeo. Or even my fifth, or tenth. I know that impressive first deployments don’t always translate to long-term success. So, what is the real potential of something like Chaotan one? To answer, let’s dig into what supercritical CO2 actually is, how it measures up to its steamy competition, and whether engineering challenges might keep it from living up to the hype.

What Is Supercritical CO2?

Right now, steam power is responsible for 80% of the worlds electricity. ⁴ One of the great things about supercritical CO2 generators is that they can be used to upgrade existing power plants. We don’t have to build whole new facilities, but instead just swap in a more efficient generator. That’s a big deal.

So, what is supercritical CO2? We all know carbon dioxide is a gas at normal temperature and pressure. When you heat it above 31 degrees Celsius and compress it above 7.39 MPa, which is about the pressure you’d feel half a mile under the ocean’s surface, it enters what’s called a supercritical state. ⁵⁶

This supercritical state is super weird. It’s more Florida than Ohio. Supercritical CO2 is somewhere between a gas and a liquid. It has the density of a liquid with the flow properties of a gas. In fact, sCO2 is more dense than steam so it can spin a turbine more effectively, meaning smaller turbines can be used. That’s wild. Chaotan One’s chief designer, Huang Yanping, put it this way:

“It’s like a strong man riding a bicycle coated with lubricating oil, allowing him to pedal effortlessly over long distances.” ⁷

I feel like they might want to work on that branding. The imagery is a little bit busy for my taste. But the point behind this picture is important. Supercritical CO2 gives you the power of a dense liquid with the low friction of a gas. And this unlikely combination is incredibly useful.

I know what some of you are thinking. Isn’t carbon dioxide bad? Well, yes, when it’s released into the atmosphere in an uncontrolled way. But in an sCO2 generator, the CO2 stays inside a closed system, so it’s recycled back through and never released. ⁸⁹ It’s only used as a contained working fluid. What’s more, CO2 is abundant. In fact we have way too much of the stuff! And it can be sourced from anywhere. Steam generators, by contrast, need to be built near a water source.

Why sCO2 Could Replace Steam

So sCO2 isn’t bad for the environment. But neither is clean, delicious steam. So why might sCO2 have an edge over steam generators that have dominated for over 100 years?¹⁰

The reason is that sCO2 can do the same job more efficiently in a smaller package. Both steam and sCO2 generators work by coupling with a heat source, like a nuclear reactor, an industrial process, or concentrated solar. Steam generators use that heat to turn liquid water into steam, which uses the force of its expansion to spin a turbine to generate electricity. Think of how a baker might use excess heat from an oven to help bread rise faster. The excess heat gets put to work rather than going to waste. ⁹¹¹¹²

The specifics matter. Steam generators use what’s called the Rankine Cycle, while sCO2 generators use the Brayton Cycle. The key difference? sCO2 Brayton cycles can theoretically hit efficiencies of 50% or higher, compared to the Rankine cycle’s roughly 33%. ⁸ To put that into perspective, Sandia Labs engineer Darryn Fleming notes that just a one percent improvement in power plant efficiency “translates into millions and millions of dollars because less fuel is burned to make the same amount of electricity.” Remember, 80% of the world’s energy is generated by steam. ¹³ A potential jump from 33% to 50% isn’t incremental. That’s transformative.

But a quick aside … how exactly do these cycles work?

Rankine Cycles

Steam powered generators utilizing the Rankine cycle have been in use for over 100 years. ¹⁰ The Rankine cycle has four main components: a boiler (heat exchanger), steam turbine, condenser, and a pump. A Rankine cycle needs heat input, usually from a boiler from burning fossil fuels or recovering heat from nuclear reactors or industrial processes. Rankine cycles are a closed loop, meaning that the water inside the system is recycled back through and used again. ¹²

The first step in the Rankine cycle is pumping liquid water into the boiler, where it is heated into a superheated vapor. Then, the steam moves into the turbine, where it is expanded to do the work of moving the turbine. After the steam transfers its energy to the mechanical work of turning the turbine, it flows through to the condenser, where it is cooled back down into liquid water. Then, the pump puts the water back into the boiler to start the process over again. The work done to the turbine by the steam turns a shaft of an electrical generator, producing the electricity. The whole process is a kind of thermodynamic rinse and repeat. ¹²

Brayton Cycles

So what’s different about Brayton cycles? With a Brayton cycle, you’re usually cooking with gas: not water, then vapor, then water again. But with sCO2, you’re starting with a gas that has been turned into something else entirely, which we’ll get to in a bit. Brayton cycles can be much more efficient than Rankine cycles because they operate at temperatures way hotter than stream, and because they don’t lose energy in the phase change between water and vapor.

There are two main types of Brayton cycles: directly heated and indirectly heated. Brayton cycles that are coupled with industrial processes and other heat recapture sources like nuclear, solar, and geothermal are indirectly heated cycles. ⁹¹¹ We’ll focus on this type.

Indirectly heated Brayton cycles are a closed system that has very little CO2 loss or addition once the system is charged with the CO2 needed to start the process. Indirectly heated Brayton cycles have some different components compared to the Rankine cycle. Brayton cycles need a heat exchanger and turbine just like the Rankine cycle. Instead of a condenser and pump, Brayton cycles need compressors and heat recuperators. ⁹¹¹¹²

Indirectly heated sCO2 Brayton cycles begin by using the heat exchanger and compressor to get the CO2 into supercritical state, then the sCO2 is moved through a turbine to collect energy to generate electricity. After the energy is moved through the turbine, the CO2 goes through the recuperators and back to the compressor, ready to start the cycle again. ⁹¹¹ The recuperators are an important part of the process because they increase efficiency. They capture excess heat and use it to preheat the compressed CO2 going back to the start of the cycle. This keeps the cycle going with minimal energy wasted through heat. ⁹¹¹

So again, that brings us back to why sCO2’s potential jump from 33% to 50% efficiency is so transformative.

On top of the efficiency gains, sCO2 systems are smaller, don’t need water for cooling, and can operate in arid climates.¹¹ But can it actually deliver? Let’s look at who’s building these things.

China’s Chaotan One

China has been the first to commercialize sCO2 power. But the path there was far from easy.

It started in 2009, when Huang Yanping, now CNNC’s Chief Scientist and Chaotan One’s chief designer, received a handwritten note at a symposium from senior nuclear scientist Academician Sun Yufa. The note suggested supercritical CO2 had research potential. Huang later recalled:

“The slip was light, but it felt heavy in my hand. It was not an order, but rather a possible path to the future, given by a senior scientist based on deep scholarship.” ¹⁴

Now my notes at a conference might lead to a doodle. This note kicked off a 17-year development program at the Nuclear Power Institute of China. The team faced constant challenges. When they realized they needed precision heat exchangers using a specialized welding technique, they approached the world’s leading manufacturer for access. But they were refused entry to the workshop … four separate times. The countries that could perform this vacuum weld were forbidden from exporting the technology to China. Huang’s response?

“This actually made us completely clear-headed. We resolved to do it ourselves.”¹⁴

To get their Brayton cycle running, they had to activate a brazen cycle. What followed was 829 days of welding development, with 218 parameters tested, 27 rounds of optimization, and 49 process tests. But the work paid off. The breakthrough came in winter 2021, when the welding process was finally successful. ¹⁵

Chaotan One began commercial operation on December 20, 2025, at the Shougang Shuicheng Steel foundry. It runs on waste heat from the steel foundry, which was heat that was previously just … wasted. CNNC claims an 85% efficiency increase and 50% more power output. ²¹⁶ Those are impressive numbers. The super critical question for the future of super critical CO2 is whether this performance can hold up over years of continuous operation. More on that in a minute.

The US Approach: Sandia’s STEP Program

China isn’t alone in pursuing this technology. The US has been working on sCO2 power for even longer … but with a very different philosophy.

Sandia National Labs in New Mexico has been at this since the late 2000s. Their founding researcher, Steve Wright, built the first US sCO2 test loop and offered a description I love:

“This machine is basically a jet engine running on a hot liquid.” ¹⁷

That’s … not a bad way to think about it. Probably a better image than the oily buff guy winning the tour de France.

The program’s most dramatic moment came in April 2022, when Darryn Fleming, who’s spent over 14 years on sCO2 Brayton cycles, led his team in delivering sCO2-generated electricity to the grid for the first time. Ten kilowatts for 50 continuous minutes. ⁸ And here’s the kind of detail I love. Fleming figured out that commercial elevator power electronics used permanent magnet rotors similar to their equipment. To feed power into the grid, you need electronics that can convert a high-speed turbine’s output into synchronized 60-hertz AC. Elevator drives already did exactly that for the same type of permanent magnet motors Fleming’s turbine used. That cross-industry insight helped to crack the grid synchronization problem they’d been stuck on.⁸¹⁷

Rodney Keith, Sandia’s Advanced Nuclear Concepts manager, put that milestone in perspective:

“Maybe it’s just a pontoon bridge, but it’s definitely a bridge.” ⁸

Since then, the STEP Demo project has generated 4 megawatts of grid-synchronized power in October 2024. ¹⁸ They’re targeting commercial designs by the mid 2030s. ¹⁹

If it works, why hasn’t the US pushed to commercialize it? It comes down to philosophy. The US tends toward methodical, long-term testing. Basically, working out the kinks before scaling up. ²⁰ China follows a different approach based on the idiom “crossing the river by feeling the stones.” ⁶²¹²⁰ Build it, deploy it commercially, and solve problems as they come. If it works, scale fast. If it doesn’t, cut your losses. Building Chaotan One is “feeling the stone” and if it holds up, they’ll keep crossing the river.

Here’s What They’re Not Telling You

Does a generator technology with high efficiency, a tiny footprint, productive use of CO2, AND the ability to run in dry climates seem too good to be true? Well … it might be.

CleanTechnica analyst Michael Barnard put it bluntly: “A system that starts at 15 MW and delivers 13 MW after several years with rising maintenance costs is not a breakthrough. It is an expensive way to recover waste heat.” ⁶ That’s a fair challenge, and it deserves an honest look.

The thing about sCO2 systems is that most problems don’t show up as dramatic failures. They creep in as efficiency losses that compound over time. ⁶

The precision heat exchangers that Huang’s team spent 829 days perfecting? They are an engineering feat in themselves, with precision etched microgrooves and vacuum and diffusion welded plates. They’re nearly impossible to repair. If one leaks, the whole unit gets replaced. ¹⁵⁶ Seals are equally tricky. Compare it to the A/C in your car. If there’s a bad seal, freon leaks out and the system can’t cool efficiently. Same principle here. sCO2 moves like a gas and can leak through tiny imperfections. The intense heat and pressure make those leaks more likely over time. ⁶

Corrosion is another serious concern. sCO2 is inert when dry and pure, but small amounts of contaminants can change the chemistry drastically, causing localized corrosion that shows up as sudden outages rather than gradual decline. ⁶ CO2 can even absorb into pure water, turning it into carbonic acid, which has plagued commercial heat exchangers for ages. It absolutely wreaks havoc on these precision components.

Then there’s the machinery itself. The Brayton cycle depends on smooth surfaces, so even small changes affect how the fluid flows through the compressor. Like the difference between rolling a skateboard on concrete versus gravel. Rough surfaces need more energy to maintain pressure. ⁶ Not only that, many of the components, including the bearings, must be reengineered for sCO2. Conventional bearings do not work well in this environment. While Sandia and its affiliates have designed some new foil bearings that are a little more suitable, there are many challenges like this ahead. [18]

For Chaotan One, contamination from its heat source might be the biggest near-term risk. The generator takes in steel plant exhaust, which carries waste that builds up a film on heat transfer surfaces. Kind of like hard water buildup in your shower head. This reduces efficiency over time. ⁶²²

Over 2-5 years, analysts estimate a 40-70% probability of measurable performance degradation for a system like Chaotan One. ⁶ Barnard’s direct challenge:

“If both the Chinese and U.S. installations run for five years without significant reductions in performance and without high maintenance costs, I will be surprised.” ⁶

That’s a fair challenge. And honestly? I think it’s exactly the right question to ask.

So here’s my take. Am I excited about supercritical CO2 power generation? Absolutely. The idea that you can upgrade existing facilities by swapping out the steam generator for something more efficient without building from scratch … that’s incredibly appealing at a time when we need more power, like, yesterday. I have friends whose electricity bills have jumped significantly because of data center build out madness , and the prospect of making power generation cheaper and more efficient is something I care about.

But I’m also a pragmatist. This technology has been theoretically promising since the 1960s, and it took until December 2025 for someone to build a commercial unit. I’m not sure if a technology should be collecting finite R&D dollars when it could be collecting social security. The engineering challenges are real, and we won’t know if Chaotan One truly delivers until it’s been running for years. Barnard’s five-year challenge is the test that matters.

Where I get really intrigued is the applications beyond steel foundries. Nuclear facilities with carefully controlled inputs. Concentrated solar. And here’s one that keeps nagging at me: data centers. The US is building AI data centers at a staggering pace … we’re talking about 33.5 million square feet of data centers in 2025 alone. They generate enormous waste heat, so could sCO2 generators help offset some of the rising energy demand they create?

sCO2 power won’t replace steam overnight. But if Chaotan One holds up, and if the STEP program delivers on its 2030s timeline, we might be looking at the beginning of something genuinely transformative. I’ll be keeping a close eye on this one.