Why did HPE buy Cray?
Is the tech giant branching out into new areas, or consolidating an existing strategy?
Late on Friday, Hewlett Packard Enterprise (HPE) dropped some big news in more ways than one.
Recently it has been working with the US Department of Energy on exascale computing projects, and announced in May its 'Frontier' project that it claimed will be the "world's most powerful computer" with a performance of greater than 1.5 exaflops. Although it's not due to be delivered until 2021 and a lot can happen in that time.
So what does this mean for HPE and the wider enterprise hardware market?
For a start, this is the second acquisition HPE has made in the supercomputing/HPC arena, having bought SGI in 2016, and it seems both the company's rationale and integration plans are largely the same. Cray, like SGI, provides HPC hardware that sits alongside HPE's own Apollo servers. But the similarities don't stop there. Both companies, prior to their acquisition, have been pouring resources into AI and machine learning (ML) systems; an area in which HPE is keen to expand.
Indeed, John Abbott, co-founder and distinguished analyst at 451 Research told IT Pro that while exascale is the "obvious goal that Cray will enable HPE to target", AI and ML may be the real draw.
"Cray's more recent focus on heterogeneous processors and clustering, and its interesting work on dynamic runtime software stacks that enable AI and data analytics workloads to be run on HPC-like infrastructure, will potentially be of even more interest to HPE as it looks to broaden out the applicability of supercomputing and HPC to other market areas, including hyperscale clouds," he said.
It seems HPE does have big plans for Cray spread across many of its business units, based on the company's initial announcement.
GreenLake, HPE's pay-per-use private cloud service, will in the future offer HPC-as-a-Service (HaaS? HPaaS? HPCaas? The possibilities are endless) as well as AI and ML-based analytics.
Its hardware will be separately incorporated into HPE's product line-up and the business also sees the opportunity to bolster its big data processing tech.
In short, this can be seen as a further example of the company showing its "laser focus", to use the tech world lingo, on high-performance hardware and emerging(ish) analytical systems having spent the past four years following its split from HP Inc spinning off non-core business units. It also shows the company is very serious about making a success of GreenLake by offering as many top-notch services as possible.
For HPE's main competitors, such as Dell Technologies, IBM and others, this acquisition may have come as something of a surprise, as it did to some industry watchers. Whether or not it will perturb them, though, is another matter.
Roy Illsley, distinguished analyst at Ovum, tells IT Pro the acquisition can be seen in one of two ways.
"Either it's a desperate attempt to create a market they can own and put their name to in an increasingly competitive market, particularly from the Chinese vendors," he said, "or it's a smart strategic move to position itself ready for the next wave of computational demand."
On this second point, however, Illsley notes that quantum computing, which is being pursued by the likes of IBM, Google and Microsoft, may render supercomputers pointless in 15 years or so. In this light, the acquisition could be "a mid-term strategic move to make cash before the next paradigm shift".
Whichever way the dice fall, more information about the acquisition and where it lies in the grand scheme of things will likely be revealed at HPE's annual Discover conference in mid-June.
Preparing for long-term remote working after COVID-19
Learn how to safely and securely enable your remote workforceDownload now
Cloud vs on-premise storage: What’s right for you?
Key considerations driving document storage decisions for businessesDownload now
Staying ahead of the game in the world of data
Create successful marketing campaigns by understanding your customers betterDownload now
Solutions that facilitate work at full speedDownload now