Carbon dioxide, used for years to force crude oil out of old wells, likely will not replace water in fracking anytime soon because of technical challenges and limited infrastructure, says General Electric Co , which is studying the issue under a $10 billion research program.

The delay means energy companies will continue to use more than 2 million gallons of water for each fracked well, equal to baths for some 40,000 people, stressing water supplies in arid American states and likely delaying fracking’s expansion to western China and other water-stressed regions.

GE, which is making a push into oilfield technology, is studying how a chilled form of CO2 known as a “super-critical fluid”—neither a liquid nor a solid—could be used as the new industry standard for hydraulic fracturing, the process commonly known as fracking.

The conglomerate is working on the project with Statoil ASA, the Norwegian oil and gas producer, as part of its ecomagination program, which also is focusing on gas turbine efficiency, wind blade design and other energy projects.

“Our ultimate vision is to have a fracking process that uses no water, but we’re a ways off from that,” Andrew Gorton, a GE mechanical engineer leading the project, said during a tour of the company’s research facilities in upstate New York.

The hydraulic fracturing of rock, or fracking, has allowed the global energy industry to access vast new supplies of oil and gas. Fracking opponents see a range of potential environmental damage from the process but are cautiously optimistic that using CO2 instead of water could reduce those risks.

CO2 fracking was used on a small scale in the 1990s by the company Canadian FracMaster before it filed for bankruptcy protection. Engineers say they want to figure out how to widely replicate the process across many different geologies.

Researchers are also trying to find the best viscosity, or thickness, for the CO2 at its chilled state to carry proppant, a type of sand that holds open cracks in rock so oil and natural gas can escape, much like water does in current methods.

“The hope is we can find a way to do it,” said Mark Little, GE’s chief technology officer.

Studies have shown wells fracked with CO2 tend to produce more oil or natural gas from the outset because CO2 fracks tend occur at a higher pressure than ones that use water.

Safely and cheaply transporting CO2, a compressible gas, on trucks to remote wells is also a concern where pipelines lag.

GE is separately studying with the U.S. Department of Energy how coal-fired power plants could best capture CO2 emissions and use the gas in fracking and other uses. GE and Statoil currently get CO2 from industrial gas suppliers such as Linde and Air Liquide.

Collecting CO2 as a power generation byproduct and using it to frack would reduce greenhouse gas emissions. But fracking would turn CO2 from a chilled fluid into a gas, and GE says it needs to devise a way to trap that gas back at the wellhead.

Ideally, a well’s owner would be able to re-use CO2 at the next well it fracks, since nearly all CO2 injected would return to the surface. By contrast, energy companies cannot re-use most of the water today used to frack, though some recycling projects are trying to address that.

For years CO2 has been injected into old, conventional wells in places like California to boost pressure and increase the amount of oil that can be pumped out. This process is much less complex than using CO2 for fracking, as it doesn’t require the CO2 to carry sand or other chemicals.

(Additional reporting by Lewis Krauskopf; Editing by Terry Wade and Andrew Hay)