HPE Introduces a Simpler, Faster Path to GenAI

A new turnkey solution from HPE with NVIDIA is changing the game for organizations that want to swiftly adopt GenAI. Creating a private cloud AI solution can be extremely complex and time-consuming because it involves integrating components including servers, high-speed networking and storage, GPUs and AI software. So, HPE launched a prepackaged solution with NVIDIA—HPE Private Cloud AI— with the main components of private cloud pre-integrated and pre-validated.

Unveiled this June at HPE Discover 2024, HPE Private Cloud AI is part of HPE's NVIDIA AI Computing by HPE collaboration, and is an integration of software, hardware, and solutions available from HPE and NVIDIA. You can test it with WWT, honored to be the first partner to have it available for customers as part of the WWT AI Proving Ground.

What's exciting about this recently released solution is its ability to greatly accelerate organizations' ability to successfully adopt AI and quickly make it productive.

The offering's simplified, plug-and-play integration eliminates so many time-consuming difficult steps, such as the lengthy process of designing a solution; identifying technologies and integrating them; testing and validation work and operating the solution in-house. (This is a key benefit as it allows clients to circumvent the skills gaps most organizations face.) Instead of all that, HPE Private Cloud AI is an appliance that you just turn on. And on top of all this, it's also much faster. HPE Private Cloud AI can be deployed in a few hours, and a pre-built digital assistant can be deployed in 30 seconds as opposed to three to six months to build an AI solution in-house.

New solution capabilities

The newly released HPE Private Cloud AI is a self-service experience that provides speed, scale, and control. Clients can start small and scale big as fast as they want while maintaining all the control. Simply put, it securely accelerates innovation while improving operational efficiencies and employee productivity. 

HPE Private Cloud AI combines NVIDIA accelerated computing, networking, and software with HPE compute, storage, and software with a self-service experience. The solution includes: 

• AI platform software
• AI frameworks and library
• AI-optimized infrastructures
• and it's delivered through HPE GreenLake Cloud

Perfect timing

And on top of all that, this solution couldn't have come at a better time. ChatGPT was the "iPhone moment" when everyone saw what GenAI could do and realized its tremendous potential. Everyone wanted it—but what does "it" mean? 

WWT, a vendor-neutral consultant with deep and wide enterprise IT experience, has been helping clients narrow down use cases and determine the state of the data and where it is related to those use cases. 

Clients are realizing it's not one size fits all. While you may need 30 racks of GPUs for large language modeling, you may just need a laptop for a chatbot; inferencing may just need three servers bringing out validated designs to take them from the use case to business outcomes.

Solutions and technology decisions come next in this process, at the "walk" phase of the "crawl, walk, run" timeline. "Walk" is where a preponderance of companies are today. As these use cases are being fleshed out and the data is being analyzed, now is the ideal time for this solution—we need a portfolio to attack these use cases.

Weighing the options

Public cloud AI offerings enable organizations to get up and running on their AI projects without building infrastructure. However, there are considerations to keep in mind, such as: compliance, security, and AI governance. Additionally, the relevant data needed for AI projects is typically stored in-house, which means moving that data to the cloud, which is time-consuming and can be expensive.

On the flip side, in favor of cloud adoption, is networking. While a single data fabric is sufficient for most needs, AI oftentimes requires three separate data fabrics. The high-speed AI fabric, in particular, is critical for optimizing AI performance. Public clouds can provide this high-speed networking relatively easily, but it is challenging to achieve the same performance levels with on-premises solutions. In simpler terms, think of data fabrics as the roads that data travels on within a network. Traditionally, one main road was enough. With AI, you need three specialized roads, one of which needs to be extremely fast to manage the heavy traffic that AI demands. Public cloud services are equipped to build these high-speed roads efficiently, whereas setting them up internally can be difficult and costly.

HPE Private Cloud AI provides the same benefits as the public cloud but without the associated risks. Many organizations are uncomfortable pushing their most sensitive data into the cloud or relying solely on public cloud AI. While developers might or might not care which approach is chosen—whether it's the public cloud, building a private cloud, or another solution—IT leaders such as CIOs and CTOs often see potential issues with a wholesale move to the public cloud. They remember the problems associated with past cloud-first approaches, where initial enthusiasm for public cloud adoption often led to subsequent migrations back to private clouds. These leaders are wary of repeating past mistakes by rushing into a full-scale adoption of public cloud AI.

So, building AI infrastructure internally is costly and time-consuming. Thus, a balanced approach is necessary, one that leverages the benefits of cloud technology while addressing the concerns related to data governance, security and cost.

In summary, while public cloud AI offerings can accelerate project initiation without the need for extensive infrastructure, careful consideration of compliance, security, and governance is vastly important. Organizations must weigh the benefits against the risks and costs of moving sensitive data to the cloud, and CIOs and CTOs must remain cautious to avoid repeating past mistakes associated with hasty cloud adoption. Balancing these factors can help organizations make informed decisions that align with their strategic objectives and operational needs.

The case for HPE Private Cloud AI

Thousands of configurations are possible in designing AI infrastructure these days, all of which depend on your particular use cases, so there is great complexity and the possibility of errors. By contrast, HPE Private Cloud AI is unique in that it is already unified and something you can just turn on and use. What's more, most organizations don't currently have the resources with the necessary skill sets to do this work; finding, training, and managing these resources is also difficult, costly and time-consuming.

Having the option of HPE Private Cloud AI gives you the benefits and capabilities of the public cloud on premises. It removes significant risks and costs from AI on the public cloud as well as the complexity, cost and time of building an infrastructure solution and hiring people to design, integrate, test, validate and run it. In closing, HPE Private Cloud AI offers options of being fully managed, partially managed with various layers of management, self-managed, or managed by partners like WWT. 

Who is HPE Private Cloud for AI right for?

HPE Private Cloud AI is a perfect solution for organizations advanced enough in their journey to consider sophisticated technologies. These organizations typically have their use cases defined, their data scientists or partners analyzing their data, and they are ready to start using GenAI. 

However, this solution is not suitable for organizations that haven't done the necessary preliminary work. Many companies start with simpler AI projects like chatbots or in-house prototypes, often hosted on public cloud platforms. When they are ready to transition these projects to full production, they will likely prefer to do so in-house, making HPE Private Cloud AI an ideal choice for this stage.

Data scientists focus on data tools rather than the underlying infrastructure. HPE Private Cloud AI provides all the necessary tools, libraries, and models along with the infrastructure, making it suitable for everyone involved. Setting up these libraries and managing the diverse skill sets required can be a significant challenge for most enterprises. This solution offers a more straightforward approach, reducing the complexity and effort involved in establishing a robust AI infrastructure.

The importance of scale

Scale is the biggest benefit. Many organizations prefer to start small. When you're ready to run, you can scale quickly. You only pay for what you need, you don't have to hoard anything or pay until you are using it, and it is economically sound. You can see results quickly utilizing this solution, which will earn you more support and trust from your organization as you scale.

This solution is new and dynamic and it will keep changing, just like the market itself. This will help you future-proof your investments in AI.