This article is more than 1 year old

HPE building its 4th global 'supercomputer factory'

Facility supports a flurry of HPC development, centered in the Czech Republic

Hewlett Packard Enterprise (HPE) expanded its European footprint this week as it revealed plans for a manufacturing facility in the Czech Republic, dedicated to building high-performance compute (HPC) systems.

The new facility, located in Kutná Hora, adjacent to HPE's existing server and storage manufacturing plant and about 90km outside Prague, will be built in collaboration with Foxconn.

HPE sees the investment as an opportunity to address ongoing supply chain challenges in the region. "We are now able to manufacture the industry's leading supercomputing, HPC, and AI systems, while increasing supply chain viability and resiliency," Justin Hotard, EVP and GM of HPC and AI at HPE, said in a statement.

Slated to break ground later this year, the plant will be HPE's fourth global site dedicated to HPC manufacturing. The company says it will produce HPE Apollo systems for HPC and AI applications alongside the company's massive Cray Ex supercomputers. HPE acquired supercomputing veteran Cray in early 2019 in a deal valued at $1.3 billion after Cray experienced several years of lackluster performance.

Manufacturing the Cray systems is no small task and required HPE to spec reinforced flooring to support each Ex cabinets' 8,000-pound (3,628kg) curb weight and specialized plumbing to keep the liquid-cooled systems from overheating during testing.

HPE doubles down on HPC

HPC has become a growth driver for HPE in the years since it acquired Cray. In 2021, HPE's HPC and AI business accounted for nearly $3 billion of its $27 billion in annual revenues.

Since the acquisition, the company has inked several high-profile supercomputing contracts, including the LUMI supercomputer in Finland, the Karoline supercomputer in Czech Republic, and is slated to build France's upcoming Adastra, Jean Zay, and Maison de l'Intelligence Artificielle supercomputers.

Meanwhile, in the States, HPE, working with AMD and the US Department of Energy's Oak Ridge National Laboratory, was responsible for developing the 1.5-exaFLOPS Frontier supercomputer, which came online earlier this year.

The company also snagged a $2 billion contract from the US National Security Agency last September to provide HPC-as-a-service via the company's GreenLake platform.

First announced in late 2020, HPE's GreenLake for HPC platform saw the vendor open its entire HPC portfolio, including its Apollo systems, to customers in what it described as "supercomputing-as-a-service."

The product meant customers could deploy HPC workloads in their datacenters without having to eat the upfront cost of building and then maintaining a supercomputer. ®

More about

TIP US OFF

Send us news


Other stories you might like