Many organizations recognize AI’s transformative potential, but scaling from pilot to production remains challenging. At AI Infrastructure Field Day 3, HPE discussed how its Private Cloud for AI (PCAI) bridges this gap, offering an integrated solution that accelerates development, secures data, and drives business value.
HPE’s Turnkey AI Factory
The journey into AI requires countless architecture decisions about hardware, software, security, and integration. HPE’s Private Cloud for AI (PCAI) streamlines this process by delivering a fully integrated, pre-configured system ready for immediate use.
PCAI is an appliance that has been co-engineered with NVIDIA. It combines HPE servers, NVIDIA GPUs and switches, and powerful software stack into a single, integrated product.
The PCAI offerings come in various sizes, from a smaller developer system to large, multi-rack configurations. It is tailored for different workloads like inference, retrieval-augmented generation (RAG), and model fine-tuning.
Bridging the Gap from Pilot to Production
- AI initiatives often stall due to:
- Lack of Strategy: Projects without unified strategy planning struggle to scale.
- Security Concerns: Protecting sensitive data is critical, whether it’s code or proprietary data.
- High Costs & Complexity: Building and managing infrastructure can be costly and difficult.
HPE’s PCAI overcomes these hurdles with on-premises or co-located deployment, predictable costs (including HPE GreenLake options), and integrated tools—so your team can focus on AI that delivers business outcomes.
A Strategic Overview of the AI Factory
PCAI features HPE AI Essentials, a robust software platform built on Kubernetes. It packages leading open-source and third-party tools for analytics, machine learning, and data governance, while remaining open for custom frameworks. The built-in Data Lakehouse Gateway unifies access to all your data, streamlining workflows. Support for NVIDIA AI Enterprise (NVAIE) and NVIDIA Blueprints enables rapid development and deployment for common use cases.
The AI Factory in Action: Basketball Play Classification
To demonstrate the power of the AI Factory, speaker Mark Zyther (HPE PCI Solutions Architect) shared a compelling project he developed for a professional basketball team. The team faced two primary challenges: managing and extracting insights from a massive volume of legal documents like the NBA collective bargaining agreement and leveraging their vast video assets from games and practices.
Focusing on the video component, Zyther set out to build a model that could automatically classify basketball plays. Using an open-source model called SlowFast, he created a labeled dataset of video clips showing two types of plays: pick and rolls and isolations. He then trained the model on the PCAI platform using a Jupyter notebook.
The results were impressive. After about an hour and twenty minutes of training on a single GPU, the model could accurately identify plays in video clips it had never seen before. It was particularly good at recognizing isolations and had high accuracy with pick and rolls, especially when the play resulted in a pass.
This project showcased the entire AI development workflow on the HPE platform:
The goal of the project was to create a searchable video archive a coach could use simply asking a question. For example if he asked, “Show me every instance of Steph Curry running a pick and roll in the fourth quarter of a game in 2017,” he would instantly get the relevant clips.
This was a cool real-world demo that showed how HPE’s AI Factory can turn a complex idea into a tangible, high-value tool.
Transforming AI from Pilot to Production
HPE’s Private Cloud for AI brings together powerful hardware, flexible software, and robust security into a single turnkey solution, making it easier to move from AI pilot to real-world impact.
For an in-depth look at these solutions, watch the full videos on the AI Field Day 3 webpage.
