What HPE’s results say about the direction of enterprise AI

Hewlett Packard Enterprise (HPE) impressed markets last week with its latest quarterly earnings report in a major seal of approval for the firm’s AI services portfolio.

In Q4, the company generated $8.5 billion (£6.7 billion) in revenue, marking an increase of 15% year-on-year.

During the earnings call on 5 December 2024, CEO Antonio Neri pointed to “the continued adoption of GreenLake and the acceleration of our revenue growth in AI”, according to a transcript provided by Seeking Alpha.

Neri said the company is “positioned to lead the next wave of innovation” and pointed to its launch of HPE Private Cloud AI, a partnership with Deloitte for private AI deployment, and Unleash AI Partner Program as evidence.

It’s no secret that HPE has been pushing the idea of private AI for the past couple of years, and private cloud for several years before that. Yet these most recent results seem to suggest the company could be onto something.

As Citi analyst Asiya Merchant wrote in a note to clients, also published by Seeking Alpha: “While AI orders/revenues can be lumpy, we see the potential for stronger contribution from enterprise AI and sovereigns, which bodes positively for revenue momentum and margins ahead.”

Cutting through the jargon, this means that there is definite and growing interest in AI solutions from enterprises that deviate from the public cloud-based offerings we see from OpenAI, which resides entirely on Microsoft Azure infrastructure.

It’s not just enterprises that are interested in moving AI – and more traditional cloud workloads – into a private cloud infrastructure. The “sovereigns” mentioned above are what most people would call countries or territories and is a segment HPE is addressing with its sovereign cloud products.

It’s far from the only traditional enterprise hardware provider to do so, either. For example, Oracle announced a sovereign cloud region for Europe in 2023 and Rackspace did the same for the UK in 2024.

Public cloud experience leads to public AI hesitancy

One of the factors that may be playing into this apparent interest in – and growth – of private AI products is large enterprises’ experience with cloud computing.

At the beginning of the cloud computing revolution, public cloud was touted as the way forward across software delivery (SaaS), as well as platform and infrastructure provisioning (PaaS and IaaS, respectively). While the SaaS argument has been won in multiple areas – particularly CRM and HR – IaaS in particular is more nuanced.

Some enterprises, particularly but not exclusively those in regulated industries, quickly ran into problems with data sovereignty and data protection. This boils down to whether the assurances and guarantees provided by public cloud providers were enough to meet their compliance obligations or generally give them a sense of security around their data.

An issue for large businesses more broadly was the total cost of ownership (TCO) of using a public cloud service versus on-premises or private cloud. While public cloud can be more economical at the beginning of a project or deployment, as time goes on and deployments scale up, those returns diminish.

As laid out by Spiceworks, “although public cloud is inexpensive, the ownership cost can rise sharply if the public cloud is scaled for extensive usage. Mid-sized to large enterprises are more susceptible to this challenge as their operations can increase rapidly as per demand”.

These two issues may be making the same organizations think twice before diving into public AI offerings – which are, after all, based on the public cloud.

With private AI offerings promising greater data protection and security, as well as results that are more attuned to an organization’s needs, having been trained on its data, it’s easy to see why it’s an enticing prospect.

Ultimately, it will take a few years for the reality of the trend to shake out as the hype around and novelty of generative AI wears off. It’s also worth bearing in mind that some of HPE’s buoyant AI results are coming from supercomputing.

While these facilities are currently popping up like mushrooms, they are normally multi-year projects that are expected to last much longer than any traditional IT deployment.

But for now, business is booming in AI, and enterprises that were burned by the cloud are fueling it.