The number to start with isn’t the catalyst chemistry. It’s 9%. That’s how much of annual U.S. electricity generation the Electric Power Research Institute estimates data centers could consume by 2030, up from 4% of total electricity demand in 2023. Everything else here is a response to that gap.
A team led by Gang Wu, the Elvera and William R. Stuckenberg Professor in the McKelvey School of Engineering at Washington University in St. Louis, says it’s found a way to make low-temperature fuel cells better at the job. The work was published Aug. 6, 2026, in Nature Nanotechnology, with collaborators from Brookhaven National Laboratory, Lawrence Berkeley National Laboratory, Northeastern University and the University of Pittsburgh.
The pitch is straightforward. “If a data center is able to supply its electricity itself by using a fuel cell, it would directly convert hydrogen and other fuels into the electricity, reducing the burden on the energy grid,” Wu said.
The durability number is the one that matters
Skip the press-release adjectives and look at the test. The material held 85% of its performance after 150,000 severe voltage cycles. The researchers estimate that works out to roughly 25,000 hours of operation.
Voltage cycling is the abuse test for this class of catalyst, because it’s the thing that kills them. Platinum nanoparticles dissolve, migrate and grow larger during operation, and performance slides down with them. A catalyst that still has most of its output after 150,000 of those cycles is making a claim about the failure mode, not just about peak numbers on day one.
Why platinum forces an ugly compromise
Fuel cells make electricity by combining hydrogen and oxygen, producing water and heat along the way. Catalysts speed that reaction up while cutting energy losses.
Platinum is one of the most effective materials for the job. It’s also a precious metal, so the entire field is an exercise in using less of it. Break bulk platinum into nanoparticles and the exposed surface area jumps, which is how loadings get down to typically less than one quarter of a milligram per square centimeter.
But small particles are unstable particles. That’s the trade that has defined this work for years.
Platinum intermetallic catalysts have emerged more recently as an alternative to conventional platinum alloys, with better activity and stability on offer. Making them introduces its own bind. Keeping the nanoparticles small, evenly distributed and stingy with platinum generally means annealing below 700°C, and those temperatures are often too low to fully drive the shift from a disordered atomic arrangement to a highly ordered one. That ordering is what maximizes activity and durability in the first place.
The fix is the container, not the metal
Wu’s team went after the support material instead of the catalyst. They built porous, hollow carbon spheres threaded with orderly radial nanochannels, with substantial pore space and surface area, and controlled the size and volume of the pores.
Those channels act as cages. Platinum cobalt intermetallic nanoparticles stay densely packed and evenly spread, and the whole thing can be heated far past the usual ceiling without the particles clumping.
“Our strategy is using this new carbon nanostructure to synthesize platinum cobalt intermetallic nanoparticles that can reduce precious metal content and enhance activity and stability,” Wu said. “Traditionally, there would be a tradeoff between size and stability, but with the ordered carbon nanochannel host, platinum cobalt nanoparticles can be confined and remain stable at very small particle size even at high temperatures.”
How far past the ceiling? “Because of this special carbon nanostructured support, we could heat the platinum-cobalt catalyst to 1000°C, which is high enough to form a very ordered structure while still keeping the nanoparticles smaller than 5 nanometers and well spread out, even with industry-preferred high content of platinum in catalysts,” Wu said.
That’s 1000°C against a conventional 700°C, with particles still under 5 nanometers on the other side. The ordering that low-temperature annealing couldn’t finish gets finished.
The channels do a second job
An electrode isn’t only a place to park catalyst particles. Things have to move through it.
“The open channel structure also helps the ion-containing material, such as an ionomer, spread evenly and makes it easier for protons, oxygen and water to move through the electrode,” Wu continued. “As a result, the platinum cobalt nanoparticles built into this support showed best-in-class performance and long-lasting durability. Eventually, through further development and collaboration with industry partners, we’ll be able to solve the remaining catalyst problems and significantly advance fuel cell technologies for powering our future more efficiently and sustainably.”
Read the last sentence carefully. Wu is describing remaining catalyst problems and industry partners he doesn’t have yet. This is a lab result with a patent filed through the WashU Office of Technology Management, not a product with a ship date.
What would have to happen next
If the development work holds up, the same catalyst could serve transportation as well as electricity generation. For data centers specifically, the appeal is generating power on site from hydrogen or other fuels rather than pulling more from a grid that’s already straining.
The research was funded by Washington University in St. Louis.
Here’s the honest read on where this sits. Existing fuel cell catalysts still fall short on the combination of activity and durability the field has set as its target, and one paper reporting 85% retention after 150,000 cycles doesn’t change what’s installed anywhere. What it does change is the assumption that you have to pick between an ordered atomic structure and small, well-distributed particles. Wu’s group heated to 1000°C and got both. Watch for whether anyone can make those carbon spheres at a scale larger than a lab bench, because that’s the step where results like this usually stall.
