HPE Announces Q2 FY2023 Results

HPE has announced financial results for Q2 FY2023, just before HPE Discover 2023. Is there a glimmer of hope in the data, and what do the Discover announcements add to this potential for future growth?

Background

For the period Q2 FY2023, HPE revenue was up 4% year-on-year, with growth achieved in HPC & AI and Intelligent Edge divisions (yet again). The Aruba business continues to grow every quarter. In comparison, Compute was down 8% and Storage down 5% compared with the same period 12 months previously. This trend reflects a broader decline in basic infrastructure sales (servers and storage) experienced across the industry.

HPE has now reported nine consecutive quarters of growth (since Q2 FY2021), albeit with modest incremental gains (five quarters were less than 2% year-on-year). Intelligent Edge has been the only division to show constant growth over this period and has the most profit margin (26.9% compared to 15.2% for Compute, 7.9% for Storage and -0.2% for HPC/AI).

GreenLake

Although there are green shoots of improvement in this data, the overall trend (as shown in Figure 2) is not one of stellar revenue growth. The migration to the GreenLake as-a-service model certainly hasn’t reawakened the HPE business, which still seems to run on a cyclical basis. Of course, it could be argued that the transition to GreenLake was required to stem the exodus of customers to the public cloud.

Figure 2 – Quarterly Revenue and Profit/Loss

We highlighted (in this post from last year) that GreenLake didn’t seem to quite have the feature set and momentum that it needed. As we review the announcements from Discover 2023, it’s possible that some of the background work is starting to come through.

Discover 2023

Note: As we’ve highlighted before, we have not been briefed by HPE (despite assurances we would), so this information is taken from public sources, essentially the Discover presentations and HPE press releases. As a result, the detail will be missing, and there may be some inaccuracies or incorrect assumptions.

Probably the most significant single piece of news from Discover 2023 is the announcement of HPE GreenLake for Large Language Models (LLMs). HPE is essentially getting into the public cloud business, offering customers the ability to train and deploy AI through a partnership with a German AI start-up called Aleph Alpha. Infrastructure will be hosted in partner data centres based on HPE Cray XD supercomputers. HPE acquired Cray Inc back in September 2019 for $1.4 billion.

As yet, there is no information on the cost of the solution (or the charging mechanism), although HPE has highlighted the goal of making the GreenLake LLM service as “green” as possible. The first instantiation will run at QScale’s Canadian data centre, which uses 99.5% renewable power (HPE was the first tenant of this data centre). HPE has introduced a new Sustainability dashboard to GreenLake, so LLM customers will be able to see exactly how efficient their model training and inference processing is.

There’s currently no detail within the GreenLake for LLM service description that explains how customers will get their data to the supercomputer platforms or how multiple customers can expect to use an on-demand service without some degree of data management. As usual, the practicalities of this kind of service are the detail customers will need to understand before diving in.

Private Cloud

HPE also announced a partnership with Equinix to deliver GreenLake for Private Cloud Business Edition (who names these products??). This solution appears to be a private cloud infrastructure hosted in one of seven Equinix global data centres (initially Frankfurt, London, California, Singapore, Sydney, Toronto, and Washington DC) running virtual servers – presumably based on VMware software.

Last year HPE announced a similar solution called HPE GreenLake for Private Cloud Enterprise (another mouthful of a name). This solution simply looked to be managed servers and storage with virtualisation software and nothing that other vendors weren’t already doing. It was already possible to build a private cloud on Equinix Metal running VMware vSphere, for example.

However, with the recent acquisition of OpsRamp, HPE has the ability to deliver significant value-add over a basic virtualisation managed service. The demonstrations at Discover show the ability to manage large-scale hybrid infrastructure using the OpsRamp technology, including root-cause diagnosis and some interesting visualisation.

The Architect’s View®

As discussed earlier, in the past, we’ve struggled to see the value of the GreenLake platform, which initially appeared simply to be a repackaging of existing solutions sold with lease-based financial terms. Without some value-add, what was HPE bringing to the party?

The move to being a public cloud service provider (with LLM models and Private Cloud Business Edition) signifies a change in direction, enabling customers to consume HPE solutions without deploying on-premises. This evolution is where things start to get interesting. On-premises deployments have significant inertia and still require customers to commit for the long term. Genuine public cloud services offer more flexibility, especially for customers wanting to trade up one level of infrastructure for another without a big on-premises rip and replace.

It’s taken quite some time, but perhaps there are some positives to the GreenLake portfolio starting to come through. At some point, HPE may need to consider restructuring the business to have a more explicit reporting structure for GreenLake solutions that de-emphasises the legacy compute and storage divisions. This move may be a bit premature, but when it happens, we’ll know that GreenLake is finally a success.

This post is only available to Individual, Vendor or Enterprise subscribers. Restrictions on distribution are based on those licensing terms. Check our Terms of Service for details.