HPE debuts private AI cloud powered by Nvidia

• HPE is the latest company to tap Nvidia for an artificial intelligence partnership

• The new portfolio notably includes a private cloud platform to help enterprises easily deploy and use AI in their environments

• Its part of a broader trend toward private cloud use for AI

Hewlett Packard Enterprise (HPE) and Nvidia just teamed up to give enterprises a private cloud for all their artificial intelligence (AI) ambitions.

A new product line, NVIDIA AI Computing by HPE, merges Nvidia’s AI software and hardware with HPE’s cloud services to help businesses more easily deploy and use AI. A key part of the new portfolio is HPE Private Cloud AI, which integrates NVIDIA AI computing, networking and software with HPE’s AI storage, compute and the HPE GreenLake cloud.

During a presentation, Neil MacDonald, EVP of HPE’s Compute business, stressed that the HPE private cloud AI is not a reference architecture that would place the burden on the customer to assemble the AI infrastructure by “cobbling together piece parts, whether those are GPUs, or pieces of software or different connectivity.” Instead, he said HPE and Nvidia “have done the hard work for customers by co-developing a turnkey AI private cloud.”

HPE Private Cloud AI aims to create a user-friendly cloud experience, facilitated by HPE’s acquired OpsRamp AI copilot that helps manage IT tasks. OpsRamp IT operations are integrated with HPE GreenLake cloud for observability and AIOps across all HPE products and services. The integration also includes support for the NVIDIA AI Enterprise software platform, which speeds up data processing and simplifies the development and deployment of AI applications.

The platform comes in four sizes of configurations, all of which provide data privacy and security, and include support for inference, fine-tuning and retrieval-augmented generation (RAG) AI workloads that use proprietary data.

MacDonald highlighted that the target customers for HPE Private Cloud AI are enterprises looking to adopt generative AI (GenAI) without the high costs associated with running workloads in public clouds. He said that managing AI workloads, which are the “most data- and compute- intensive workload of our generation,” in a private cloud also helps maintain data governance and IP protection.

The perks of private AI

NVIDIA AI Computing by HPE represents the expansion of a decades-long partnership between the two companies. It’s also part of a broader trend in the industry toward using private clouds for AI (just this month, Apple presented its customers with an option for personalized AI in a private cloud).

For its part, Nvidia has become something of an AI arms dealer, putting its systems in the hands of companies that will build them into broader solutions in the market.

For example, Dell last year announced a platform for GenAI with NVIDIA AI Enterprise and VMware Cloud Foundation software, and more recently, its AI Factory. And earlier this year, Equinix announced its own Private AI platform using NVIDIA DGX. 

Melanie Posey, research director at S&P Global Market Intelligence, said these types of portfolios are becoming a must-have for many tech vendors. There are “subtle” differences in how they put them together, often related to the portfolio items they have on the shelf already.

There are plenty of reasons that an enterprise might want to have on-premises private cloud infrastructure, she added, as sharing infrastructure can come with performance and security challenges.

On that front, Nvidia’s Manuvir Das said that AI workloads often need to access the private data of a company. “Would you prefer to send that data into the cloud or keep under your control,” he added.

The private cloud also presents a cost-savings play. If a company can stand up its own infrastructure, Manuvir said, they can get faster total cost of ownership. “Data has latency to it,” he continued, “If you have petabytes, do you want to move data to compute or compute to data?”

GreenLake foundation

HPE Private Cloud AI includes a fully integrated AI infrastructure stack with NVIDIA Spectrum-X Ethernet networking, HPE GreenLake for File Storage and HPE ProLiant servers supporting the latest Nvidia graphical processing units (GPUs).

Fidelma Russo, HPE’s chief technology officer, explained that HPE Private Cloud AI using software foundation as HPE GreenLake ensures integration and tuning for AI workloads using HPE’s and Nvidia’s software.

With the HPE GreenLake cloud, enterprises can start with a few small model inferencing pilots and then scale to multiple use cases with higher throughputs. The cloud platform offers a “range of support,” from self-managed to delivering it as a fully-managed service.

Posey noted the private cloud control plane built into GreenLake is a way for customers to manage and orchestrate the hardware they consume in an opex model rather than a capex model. That’s because the private cloud construct on premises has a public cloud-like mode of consumption. Traditionally, opex models were only accessible through public clouds from the likes of AWS, Azure or Google.

The variation of that with private cloud is similar, but the infrastructure is on premises, and the benefit is that it’s dedicated to an enterprise’s environment alone.

HPE Private Cloud AI is expected to be generally available in the fall. Global IT services firms like Deloitte, HCLTech, Infosys, TCS and WIPRO have announced their support for the NVIDIA AI Computing by HPE portfolio.