HPE ushered in the agentic AI era with a new HPE GreenLake Intelligence agentic AI framework aimed at forever changing IT operations with autonomous, self-learning agentic AI assistants that ‘fix’ IT issues faster across the full multi-cloud, multi-vendor hybrid cloud stack.
“Ultimately, the goal here is being able to reason and understand in a multi-vendor, multi-cloud, full-stack context to be able to deliver the intelligence that is needed,” said Kunaparaju in a press conference unveiling the new platform. “The unique, highly differentiated HPE IP in this area is not only being able to deliver this across the HPE stack but being able to address the other vendors’ and multi-cloud estates of every enterprise application.”
In fact, GreenLake Intelligence captures “live telemetry” from customer IT environments and combines it with HPE’s “deep domain knowledge across compute, networking, storage and the full software application stack,” said Kunaparaju.
“Unless one understands the full stack from application all the way to the infrastructure, then the ability to really do this intelligence does not exist,” he said. “That is where we are bringing individual domain-specific, agentic protocol-driven outcomes combined with the intelligence that we are bringing across multiple domains.”
The GreenLake Intelligence platform is built to assist IT teams at every stage of IT operations from “simplified on-boarding” and orchestration to IT optimization, said Kunaparaju.
“The goal ultimately is very simple,” he said. “Our design with GreenLake Intelligence is to help customers fix issues faster, prevent failures earlier, shift focus to innovation and ultimately to deliver business outcomes that [IT teams] are expected to deliver.”
HPE is leveraging the orchestration automation platform that it gained from its acquisition of Morpheus in August 2024 and the AIOps platform it gained from its purchase of OpsRamp in March 2023.
“All those technologies are multi-vendor, multi-cloud assets where we have real data,” said Kunaparaju, the former founder and CEO of OpsRamp who joined HPE with the acquisition of OpsRamp. “Today we collect about 20 trillion metrics across the globe.”
GreenLake Intelligence thus is being powered by the “raw data that we collect, and we basically analyze the data utilizing all the advances that took place with GenAI and more importantly customizable foundational models that we fine-tuned,” said Kunaparaju.
Ultimately, the GreenLake Intelligence platform will result in “faster detection and remediation” of IT issues powered by “insights that are difficult or impossible for humans to extract on their own,” said Kunaparaju.
Among the use cases cited by HPE is the use of GreenLake Intelligence to make “judgement calls” to optimize workload cost and performance, said Kunaparaju.
HPE is demonstrating an operations command center capability that leverages “full-stack observability data” to determine the precise root cause of an IT issue—essentially a ‘needle in a haystack’—and then follow it up with actual remediation” of the problem, said Kunaparaju.
“These outcomes ultimately drive the productivity, the efficiency and the ability to make the enterprise act and give the business outcomes the lines of business are expecting from IT,” said Kunaparaju. “This is not a black box in terms of GreenLake Intelligence. This is accountable, expandable and explainable reasoning that [IT] teams can trust and deliver.”
A New Agentic AI-Powered HPE Aruba Networking Central
HPE has added new agentic AI mesh technology to its HPE Aruba Networking Central management system to provide what it calls “precise root-cause analysis and guided or automated remediation for complex network and security issues.”
HPE Aruba Networking Senior Vice President and Chief Product Officer David Hughes said GreenLake Intelligence brings the “next big wave” in AI to Aruba Central with an “autonomous supervisory AI module” talking to many different specialized subagents to remediate network issues.
“Unlike typical generative AI, instead of just giving you a list of things to go and investigate it can actually work its way down that list, take those actions, collect that data, analyze the overall result, come up with a root cause and from there figure out a proposed recommendation,” said Hughes. “It’s like having a teammate that can work on problems while you’re asleep and when you arrive in the morning have those proposed answers there complete with chain-of-thought logic explaining how they got to their conclusions.”
The foundation of GreenLake Intelligence is an “autonomous multi-agent orchestrator” that essentially acts as a “large reasoning model” that connects in a “mesh” to a host of other agents, said Hughes.
In the demonstration of GreenLake Intelligence at HPE Discover, there are more than 15 specialized agents tracking both real-time data and support cases or technical publications, said Hughes.
“The multi-agent orchestrator is able to draw on advice and knowledge from all those sub-agents and tie it together,” said Hughes. “The multi-agent orchestrator is autonomous. … But it can communicate with a network admin through a copilot interface. So we can have the network admin interact with the autonomous orchestrator in order to keep a human in the loop or to provide more context.”
As to a specific example of GreenLake Intelligence in a networking scenario, Hughes cited a case where wireless clients were experiencing “authentication” failures when roaming.
“You may have hundreds of these types of alerts, and a huge challenge for IT is figuring out which ones are critical and then finding time to dig down and understand what is going on,” he said.
The multi-agent orchestrator picks up on the authentication failure trigger and does a “blast rate radius analysis” to understand how big the problem is, said Hughes. “In this case it identifies 52 clients experiencing issues across two access points,” he said.
The orchestrator then notices that many of the devices are not compatible with 802.11r (fast transition enabled), identifies those devices and works out the “pros and cons” of making changes to resolve the issue, said Hughes. Ultimately, the multi-agent orchestrator determines the best course of action is to “disable the feature” even with the resulting increased network latency, he said.
“It makes that recommendation so the human copilot can adopt this immediately, schedule it for a date or have it happen during the next maintenance window,” said Hughes.
“It’s a real game-changer because, as you can see, it takes what would take a human hours going through all of this and figuring out this correlation,” said Hughes.
Hughes stressed that the GreenLake Intelligence fits broadly across the full IT stack in storage and compute.
“What’s really exciting is we can actually do it at the next level, where we are looking across all of the IT estate—compute, storage and networking—with multi-agent orchestration that is bringing in data from each of the different domains, each of which in turn is benefiting from all of these very specialized subagents,” said Hughes. “So we see the change here as being really, really dramatic, making a huge difference on the workload for IT. It includes explainability, making it easier for humans to adopt and trust the recommendations. So going forward you’ll see this GreenLake Intelligence permeating throughout our whole stack.”
The ‘Only Vendor’ That Covers The ‘Entire IT Estate’
HPE’s singular strength in networking, compute and storage sets it apart as the “only vendor that can deliver the entire IT estate with an agentic framework,” said HPE Vice President of AI, Security and Networking Product Marketing Larry Lunetta.
“Other vendors are going to talk about their solutions in a variety of ways,” said Lunetta. “What is profound about GreenLake Intelligence is we are the only vendor that can deliver the entire IT estate with an agentic framework. If you look at all the other vendors, they are missing networking. They are missing servers. They are missing storage. We have it all. That is where the breakthrough is. That is where this is going to really make the difference in the industry.”
HPE rivals coming to the agentic AI opportunity without that full IT estate stack are destined to provide a “suboptimal” offering, said Lunetta.
“When you have a poorly performing AI solution the question is: Is it the server? Is it the storage? Is it the network? Unless you can look across that [stack] and feel comfortable with the diagnosis that is being done in each of those areas you are not going to get the full benefit of agentic AI.”
HPE’s ability to deliver its agentic AI offering right now even as other vendors move quickly to bring an agentic AI offering to the market is also a “big difference” for HPE, said Lunetta.
“With all due respect to all the other vendors, if you haven’t built the data, the data science and the visibility over time it is going to take time even for big vendors to deliver this kind of capability,” he said. “It’s here and now for us because of the journey we have been on. And it is part of the standard [Aruba] Central license. We don’t charge extra for helping customers be more efficient.”
Reducing Data Center Costs By 70 Percent With Agentic AI
HPE believes it can reduce data center costs by as much as 70 percent by leveraging GreenLake Intelligence and a number of recent HPE “innovations,” said Lunetta.
Besides GreenLake Intelligence, HPE Aruba recently introduced its next-generation CX 10000 smart switches—including the CX 10040—which offload firewalls, network- attached storage and other network services with an AMD data processing unit that accelerates services inside the data center, said Lunetta. “That not only makes the servers more efficient; it also eliminates the need for associated appliances,” he said. “Because we are doing east-west firewalling, we’re cutting down the number of appliances you need.”
HPE also earlier this year introduced Morpheus VM Essentials, which provides dramatic cost savings over Broadcom-VMware Private Cloud Foundation. In fact, HPE partners have maintained that the Morpheus VM Essentials provides 10X cost savings versus Broadcom-VMware Private Cloud Foundation.
HPE is providing partners with a full arsenal of innovative offerings to hit that 70 percent reduction in costs target not just in the data center but also with campus networks, said Lunetta.
In fact, he said, HPE is going into a number of customers to reduce switching layers from three to two to even reduce costs for smaller customers.
A Network Architecture For The Agentic AI Era
HPE’s Aruba networking platform is the only network architecture that can be deployed as a SaaS application, a virtual private cloud application or on-premises air-gapped from the rest of the network, said Lunetta.
“This is the only network management solution that has these kinds of deployment options,” said Lunetta in an interview with CRN. “If you look at other choices, you can have an on-premises option that will be one architecture, you can have another architecture that is in the cloud and those data lakes are separate. They are not combined. They are not looking at all the different ways a customer can deploy a network in one place. That I think is the crucial difference [between HPE and competitors].”
HPE Aruba leverages its single data lake to find and fix problems 80 percent faster than one could do manually, said Lunetta. Furthermore, he said, HPE leverages the data lake capabilities to optimize the network. That capability is recommending best practice changes with the architecture, deployment and number of devices to improve network performance. “Think about that, you can change a setting and get 25 percent more performance without spending a nickel on additional hardware or anything else. But you can’t do that effectively if you have two different AI systems and two different data lakes because then you are not looking across the entire deployment architecture.”
HPE Alletra Storage MP X10000 Provides Agentic AI Powered Storage
HPE is extending its AI agentic capabilities into the Alletra Storage MP X10000 storage platform with support for model context protocol (MCP) servers natively built into the Alletra MP X10000.
The new platform “reimagines what storage can be” in the agentic AI era, said HPE Senior Vice President and General Manager of Private Cloud and Flex Solutions Cheri Williams.
The new agentic AI-powered storage capability delivers an “intelligent, unified data layer purpose-built for agentic AI,” said Williams.
The new Alletra MP X10000 combines block, file and object storage with “AI data and vector data capabilities,” said Williams. “Alletra MP provides a unified architecture that sets a new standard for AI data. It delivers extreme performance at scale, real-time meta-data enrichment to transform unstructured data into AI-ready assets.”
The MCP server that feeds GreenLake Intelligence opens the door to integrate the AI-ready data into any agent orchestrating business logic. “This is how you can now talk to your data with HPE Alletra using an agentic copilot in real time,” said Williams.
An ‘Astronomical’ Opportunity For Partners
The new GreenLake Intelligence platform opens up an “astronomical” opportunity for solution providers looking to deliver game-changing business outcomes for customers, said Williams.
One example of the power of agentic AI to solve issues that would be nearly impossible to take on without agentic AI is a case in which HPE software partner Aible AI built an agent model aimed at preventing animal and agricultural diseases in the state of Nebraska, said Williams.
Working with HPE Private Cloud AI, Aible provided the state of Nebraska with an analysis of diseases in agricultural scenarios including pig farming, said Williams. “Aible was able to look at behaviors of pigs and then use that for predictive analysis of potential infections,” said Williams.
In fact, within 90 minutes, Aible was able to get a model up on HPE Private Cloud AI that generated 35 agents, said Williams. That opened up a whole wave of possibilities for the state to improve agriculture issues. “The light bulb went off all over the place for these users with the art of the possible,” she said.
With GreenLake Intelligence being brought to market, the opportunity for how partners can make the lives of their customers better is “phenomenal,” said Williams.
“We are building all this out in real time, bringing these pieces together,” she said. “The power of this is the partners have access to all this technology and capability that they can go leverage to build out solutions for their customers. All the pieces are there.”
A New Cloud Operations Software Suite For The Agentic AI Era
HPE’s new CloudOps Software Suite brings together OpsRamp, HPE Morpheus Enterprise Software and HPE Zerto Software in a simple-to-consume software suite.
HPE said the new integrated software suite provides automation, orchestration, governance, data mobility, data protection and cyber resiliency across multi-vendor, multi-cloud, multi-workload infrastructure.
As its steps up its multi-vendor and multi-cloud orchestration, network management and AIOps suite offensive, HPE is also providing customers a new HPE Cloud Commit software buying capability that provides discounts for locking in multi-year cloud software purchases similar to the cloud credits offered by public cloud hyperscalers like Amazon Web Services.
Customers can aggregate their existing HPE GreenLake spending with future commitments across the breadth and depth of HPE’s software offering from the Cloud Software Suite to storage solutions like the Alletra MP B10000 and 5000 to HPE Private Cloud AI and HPE Private Cloud Business Edition and HPE Morpheus VM Essentials.
The HPE Cloud Commit program provides customers with the flexibility to easily implement the software they require on an as-needed basis through the cloud credits. “You don’t know exactly how much you are going to need of each of these software components,” said HPE’s Williams. “So you make the commit and then you have the credits, adding more of this or that, depending on what you need as you evolve your operating model.”
HPE said it is building out a plan over the next six to nine months for partners to participate in the HPE Cloud Commit program.
A ‘Seminal’ Moment For HPE: The Only Vendor With The ‘Full Complement of AI Factory Solutions’
GreenLake Intelligence builds on all the robust innovation that HPE has built into GreenLake over the last four years and “catapults it into the agentic AI world,” said Williams.
“This is a seminal moment for HPE, absolutely a seminal moment,” said Williams. “HPE is the only vendor in the industry that can offer the full complement of AI factory solutions including agentic AI from an enterprise turnkey appliance to fully customizable AI factories that include the Cloud Ops Software Suite running on GreenLake with GreenLake Intelligence. There is nobody else that can put that entire complement together other than HPE.”
Williams said the GreenLake Intelligence Framework is a tribute to HPE President and CEO Antonio Neri’s GreenLake hybrid cloud vision that was put in place when he became CEO seven years ago and is now a full multi-cloud, multi-vendor hybrid platform for the agentic AI era.
“At HPE Discover for several years we have talked about AI and hybrid and network connectivity,” said Williams. “We have it all. It has all now been brought together through the combination and integration into the strategy and all of it tied together with GreenLake.”
