Hewlett Packard Enterprise (HPE) is ramping up its artificial intelligence (AI) initiatives with some exciting new plans announced at the HPE Discover Barcelona 2023 event.
One key update is HPE’s expanded partnership with Nvidia, which will enhance both hardware and software to better support AI for enterprise tasks. The HPE Machine Learning Development Environment (MLDE), initially launched in 2022, is getting new features to help businesses create, customize, and use AI models more effectively. Plus, MLDE will now be offered as a managed service on AWS and Google Cloud. HPE is also bolstering its cloud services with new AI-tailored instances in HPE GreenLake and improved file storage performance to boost AI workloads.
These enhancements are all part of HPE’s strategy to build a comprehensive AI-native architecture optimized from hardware to software. Modern AI tasks require significant computational power, vast amounts of data, and large-scale processing capabilities.
Evan Sparks, VP and GM of AI Solutions and Supercomputing Cloud at HPE, highlighted that AI workloads need a fundamentally different architecture compared to traditional computing tasks like transaction processing and web services, which have dominated the industry for years.
With the updated HPE MLDE, the goal is to help businesses integrate AI into their operations more seamlessly. Sparks mentioned that new features would assist customers in adopting generative AI workflows, including tasks like prompt engineering, retrieval augmented generation (RAG), and fine-tuning pre-trained models. This will help businesses implement advanced AI features they hear about from top research labs into their own systems.
The HPE Ezmeral Unified Analytics software suite is also receiving upgrades, improving model training and optimization through deeper integration with MLDE. The aim here is to speed up the deployment of AI in organizations, helping them realize value faster.
For AI to be genuinely beneficial for businesses, they need to be able to utilize their own data effectively. This is where the enhancements to HPE GreenLake for File Storage come in. Patrick Osborne, SVP and GM of HPE Storage, announced significant improvements in performance, density, and throughput to handle demanding AI workloads. This includes a 1.8x capacity expansion, with further growth planned to support up to 250 petabytes of data, addressing the needs of organizations creating massive language models.
Moreover, HPE is enhancing its partnership with Nvidia to provide new integrated hardware solutions. Initially launched in June, this collaboration is expanding to support a broader range of AI tasks, including training. Neil MacDonald, EVP and GM at HPE Compute, explained that most companies won’t develop their foundational models but will deploy pre-developed models into their systems to revamp their business processes. The challenge lies in building and implementing the infrastructure needed for fine-tuning, experimentation, and deployment.
New HPE systems designed specifically for AI include the HPE ProLiant Compute DL380a, featuring Nvidia L40S GPUs, Nvidia BlueField-3 DPUs, and Nvidia Spectrum-X Ethernet technology. HPE MLDE and Ezmeral Software will be optimized for Nvidia GPUs, and HPE will collaborate with Nvidia AI Enterprise and the NeMo framework software to meet enterprise needs. MacDonald emphasized that businesses must adapt to become AI-powered to avoid becoming obsolete.