Hewlett Packard Enterprise just unveiled the 12th generation of its ProLiant server platform, which the company says focuses on delivering security, manageability, operational and power efficiency and, of course, performance in the datacenter. Over the past couple of years, messaging and focus for datacenter infrastructure vendors have shifted almost exclusively to AI enablement and partnerships with AI market leaders. The unintentional result has been a somewhat commoditized view of these hardware platforms.
HPE is countering this trend with its messaging for ProLiant Compute Gen12, which marks HPE’s first major generation launch since 2022. The company seems intent on doubling down on its silicon and hardware design — and tying that to performance and security up the operating stack. Let’s dig into the state of the market today, what HPE has delivered in Gen12 and whether these features and capabilities are likely to resonate with enterprise IT.
Mapping The Needs Of Enterprise IT — Everything’s Different Now
Enterprise IT is stretched too thin. The needs of modern businesses, the evolving threat landscape, the diversity of their compute farms and shrinking power budgets have IT organizations scrambling. Add to that the looming impact of AI moving from the cloud to on-prem — and how to support and manage these environments — and it’s enough to send many IT executives over the edge, so to speak.
As attention-getting and intimidating as AI is, we have to keep in mind that it is only one category among many of the workloads running the modern business. Today’s virtualized and cloud-native environments support everything from desktops to databases to mission-critical applications, which means they require more relevant performance — which means not simply more cores and more memory, but the right amounts of memory, compute and acceleration, all delivered at the right time and for the right workload.
The security threats keeping CISOs up at night have also become more sophisticated. In this context, AI is both a friend and a foe. Worse yet, advances in quantum computing mean that the ability to break many of the public-key encryption algorithms that are the foundation of the most popular cryptographic systems is right around the corner.
Meanwhile, IT organizations often discuss having bespoke architectures for bespoke workloads. The discussion should be about bespoke architectures for bespoke datacenters. This implies a compute platform that can support a wide range of applications with contradicting performance characteristics running simultaneously at the server, rack and datacenter levels. Complementing this architecture should be an AI-assisted management platform that delivers automation. This platform ought to handle the tedious and mundane tasks on IT organizations’ to-do lists and free them up to focus on the big-ticket items, such as enabling AI across the business.
To sum up the problem statement, enterprise IT requires a server platform that can support any workload at any time in the most performant yet efficient manner. Oh, and this server platform must have a security profile designed for tomorrow while managing itself today. Just for kicks, throw in a cloud operating model as a way to view and monitor all of the servers spread across the globe.
Simple enough? HPE has attempted to deliver all of this and then some with its ProLiant Compute Gen12 platform.
Built-In Security And Post-Quantum Cryptography
HPE is leaning heavily into security for this release, so it’s worthwhile to set some context. Before an OS boots, the millions of lines of code that make up the OS kernel, drivers and so on run with unfettered access; as such, they are ripe targets for hackers who want to install malware in what is known as a rootkit attack. These attacks can run undetected and with full system access for months, even years.
In response to this threat, HPE deployed silicon root of trust as a key element of its Integrated Lights Out management technology in its 10th generation of ProLiant back in 2017. (More specifically, that was iLO 5.) In simple terms, silicon root of trust creates an immutable fingerprint for the boot environment that is monitored as a server boots. Any deviation during the boot process is detected, and the server reverts back to the last known good (and clean) boot image. Soon, silicon root of trust was taken up across the industry, and it became a standard capability for all server vendors. Unsurprisingly, HPE has also included it in the release of iLO 7 for the ProLiant Gen12.
Here in 2025, quantum computing is the next significant threat looming. In response, HPE has built a dedicated security processor that stores passwords and keys in a quantum-resistant fashion. In fact, the company claims to be the only major server vendor to meet the rigorous standard of the U.S. and Canadian governments’ Federal Information Processing Standard 140-3. For any company that is working with the government or in a regulated environment, FIPS certification is important. As HPE launches ProLiant Gen12, I have to assume that securing this official validation will be a high priority. (For context, note that previous generations of ProLiant have achieved FIPS 140-2 validation.)
The IT and service provider community may be more familiar with NIST — the National Institute of Standards and Technology — Cybersecurity Framework. ProLiant Gen12 also meets these standards. In fact, meeting these standards, which includes FIPS validation, has been part of ProLiant’s edict since its Gen10 release.
While this is all good, it is more important to state what this should mean for ProLiant customers: the foundation for a zero-trust architecture rooted at the lowest levels and delivered in a handshake fashion up the stack. This should give enterprise IT the confidence to run any workload, anywhere with confidence in its security posture today and tomorrow.
AI-Driven Insights And Management
HPE is also using the launch of Gen12 to promote updates to its Compute Ops Management platform. Here’s the setup: IT organizations spend an inordinate amount of time on the mundane — deploying and taking care of servers throughout their lifecycle. While there is no single study of this that I feel comfortable citing, I do know from past experience in IT that a significant chunk of an IT admin’s time — in some cases up to 50% — is spent on server monitoring, management, tuning and so on.
HPE claims that with COM, organizations can cut up to 75% of the time spent managing servers and up to $152,000 in costs (travel, software) over three years. What sticks out to me is the 75% figure. Even if the actual number turned out to be somewhat less than that, it represents a big shift from spending hours of each working day reacting to problems and routine headaches to being productive and making progress on the initiatives that really matter.
In this connection, I would love to see two things. First, I want to see HPE lean more into AIOps. Based on what I am seeing from HPE, the company can already make this claim — it just seems to be more cautious in its use of the term. However, the company already has the IP and software portfolio (including OpsRamp, Morpheus and much more) to take a market leadership position in this area. It’s a matter of integrating all of these discrete parts.
Second, I’d like to see HPE be bolder in its storytelling about datacenter management. While it might not be as cool as AI, managing the modern IT estate — from the edge to the clouds to the core datacenter — is a huge responsibility and a major focus for the typical enterprise IT organization. HPE has all the pieces to address this, and I don’t doubt it has the vision. Now I’d like to see it do more to help the market see what the future looks like and how COM and its other products and IP drive the autonomous datacenter.
The Impact Of Efficiency
The final area of focus with Gen12 has been efficiency, across two vectors. The first is efficiency from the perspective of hardware refresh. When comparing Gen10, which is still widely deployed, to Gen12, HPE claims a server consolidation ratio of 7:1, along with 65% power savings. This is impressive streamlining whether we’re talking about physical footprint, operating cost or electricity budget.
Here’s what I like about HPE’s claim. It’s based on a realistic assessment of the hardware currently sitting in the enterprise datacenter. It hasn’t taken the top bin and highest-core-count processors to deliver the most impressive consolidation ratios. It is looking realistically at what gets deployed in a datacenter to provide more real quantifications to its IT customers.
Where Is AMD?
One of ProLiant’s curious elements is its processor lineup. At launch, the only x86 processors available for Gen12 are Intel Xeon 6. The lineup today includes traditional Intel-based Xeon servers, with the blade and four-socket DL580 options shipping later. Noticeably absent from the lineup are any AMD EPYC models.
I suspect this is because Xeon 6 may have required physical changes to the motherboard for support. Xeon 6 is not pin-compatible with 5th Generation Xeon processors, meaning that a server manufacturer can’t simply put one CPU in the socket of the other.
I think the absence of EPYC is also a matter of pin compatibility. Because the 5th Generation EPYC is pin-compatible with the previous generation, HPE was able to use its ProLiant Gen11 servers to support these new processors when they launched. So while HPE customers deploying EPYC will have to wait for Gen12 support, they are able to fully leverage the capabilities of the latest EPYC CPUs using other HPE gear.
Given that security is perhaps the biggest differentiator for Gen12 and that additional security comes through iLO 7, it is important to see AMD EPYC support stated as a plan of record. I hope HPE will clarify this soon.
Has HPE Differentiated These Servers?
Has HPE done enough to differentiate Gen12 from the competition? Are servers simply commoditized at this point? Are they just vessels for the AI accelerators that sit inside? I suspect many pundits look at server infrastructure this way, but frankly, it’s a bit of a simplistic view.
In my estimate, HPE has delivered a differentiated platform from both hardware and software perspectives, as touched on above. And based on what it announced at HPE Discover in June 2024, the company also differentiated how it partnered with Nvidia on the AI front and its private cloud offerings (see my coverage here).
This is the reality of the datacenter market. Differentiating from a hardware perspective is increasingly difficult for server manufacturers. However, I believe that HPE has maximized its opportunity to differentiate ProLiant Gen12 with security and iLO 7.
Still, the biggest differentiation is in software — the operating stacks, manageability and workload enablement. Over the years, HPE has created this differentiation through in-house development and acquisition. So, when it does things like striking a partnership with Nvidia for its AI factory, the result is not just HPE hardware with Nvidia acceleration and software. It’s HPE leveraging its IP to make Nvidia AI solutions more easily deployed, consumed and managed. If I’m running an enterprise datacenter, that’s a much more appealing proposition.
The challenge for HPE is delivering a vision and message that gets this across to its partners and the IT consumer. This means cutting through AI hype and making its value known — and this is not a time and not the market for messaging modesty.
I’ll check back on ProLiant Gen12 and its market traction around the Discover 2025 event in June to see where things stand regarding the dynamics and challenges I’ve outlined here.
