Hewlett Packard Enterprise (HPE) today announced that it would acquire supercomputer company Cray for $1.3 billion in a cash offer. This is a big move by HPE to boost its position in the high-performance computing (hpc) market.
“HPC is a growing market, one of the fastest growing in the data center market,” said Patrick Moorhead, founder, president, and principal analyst at Moor Insights & Strategy. “HPE for years expressed the intent to increase its capability into [HPC].”
Cray is a supercomputer manufacturer that was founded in 1987 and headquartered in Seattle. The company’s product portfolio includes supercomputing, storage, and data analytics offerings.
By integrating its existing HPC offerings — which include HPE Apollo and SGI — with Cray, HPE plans to build an integrated platform, scale, and resources for the exascale era of HPC.
Moorhead says the success of this acquisition hinges on HPE’s ability to integrate Cray’s technology. “HPE’s made a lot of acquisitions. Some of them have gone really well, like Aruba, but there were some software acquisitions previously that didn’t go very well. So I’m really interested to see how that is going to work,” he said, adding that overall it does seem to be a good move for HPE.
“What I’m interested to see is combining HPE’s…I’ll call it scale and some of their unique consumption models like GreenLake, combined with Cray’s intellectual property, particularly along the connectivity and networking,” Moorhead said. “That’s one of the unique things that Cray brings to the table is super fast methods to connects banks and servers together so they act like one logical unit.”
And, he added, these connectivity and networking factors were one of the main reasons why Cray recently won a $600 million exascale supercomputer contract with the U.S. Department of Energy (DOE). “The safe bet with the DOE was to go with an Mellanox, Nvidia, and IBM solution but instead they did an AMD, Cray solution and I think a lot of that had to do with the networking capabilities that Cray brings to the table,” Moorhead said.
This contract involves Cray building a supercomputer for the DOE’s Oak Ridge National Laboratory that can exceed a performance of 1.5 exflops (a measurement of calculations per second). The entire exascale system for the DOE will include Cray’s Shasta supercomputer architecture, Slingshot’s interconnect network, and AMD graphics processing unit (GPU) technology. According to HPE, Cray’s portion of the contract is valued over $100 million.
For Moorhead, Cray winning this contract “proved to me that their technology is unique.”
HPE plans to leverage Cray’s technology for a number of new exascale HPC offerings. These include developing HPC as-a-service with artificial intelligence (AI) and machine learning (ML) tools for it HPE GreenLake portfolio, and building a comprehensive HPC infrastructure with compute, storage, system interconnects, and additional software and services for data-intensive use cases.
“ICray does this super high-end [computing], and I’m interested to see HPE’s scale and HPE taking Cray’s intellectual property and moving it to not the top 10 but the other 490 in the top [Fortune] 500,” said Moorhead.