By Stephen Nellis
SAN FRANCISCO (Reuters) -Nvidia and Hewlett Packard Enterprise said on Tuesday they are partnering with the Leibniz Supercomputing Centre to build a new supercomputer using Nvidia’s next-generation chips.
The Blue Lion supercomputer, as the project is called, will become available to scientists in early 2027, using Nvidia’s “Vera Rubin” chips.
The announcement, made at a supercomputing conference in Hamburg, Germany, follows Nvidia’s announcement that the Lawrence Berkeley National Lab in the United States also plans to build a system using the chips next year.
Separately, Nvidia also said that Jupiter, another supercomputer using its chips at German national research institute Forschungszentrum Julich, has officially become Europe’s fastest system.
The deals represent European institutions aiming to stay competitive against the U.S. in supercomputers used for scientific fields from biotechnology to climate research.
Long before it became an artificial intelligence powerhouse, Nvidia set out to persuade scientists to use its chips to speed up complex computer problems, such as modeling climate change. Those problems required many precise calculations that could take months at a time.
Nvidia is now working to persuade scientists to use artificial intelligence. Those AI systems can take the results of a few precise calculations and use them to make predictions that, while not as accurate as the fully calculated results, can still be useful while taking far less time.
Nvidia on Tuesday unveiled what it calls its “Climate in a Bottle” AI model. In a press briefing, Dion Harris, head of data center product marketing at Nvidia, said scientists will be able to input a few initial conditions such as sea surface temperatures and generate a forecast for 10 to 30 years in the future and see what the weather may be like at any kilometer or so of the earth’s surface.
“Researchers will use combined approach of classic physics and AI to resolve turbulent atmospheric flows,” Harris said. “This technique will allow them to analyze thousands and thousands more scenarios in greater detail than ever before.”
(Reporting by Stephen Nellis in San Francisco; Editing by Lincoln Feast.)
