•
Nvidia and VMware announce an extensive collaboration to empower enterprises with customized generative AI technology, focusing on security and privacy.
• The VMware Private AI Foundation with NVIDIA platform, set for an early release, provides access to Nvidia's accelerated computing and generative AI software for enterprise customization.
• The platform ensures secure data access, supports up to 16 GPUs for scalability, and utilizes Nvidia's NeMo framework for building generative AI models with efficient resource sharing.
In a joint announcement at VMware Explore 2023, Nvidia and VMware unveiled their extensive collaboration expansion, unveiling a groundbreaking platform designed to empower enterprises, especially the vast user base of VMware cloud services, to tailor generative AI technology to suit their unique business needs.
Customization has emerged as a pivotal focus for Nvidia, given the rapid proliferation of generative AI technology over the past ten months. This approach aims to empower enterprises to harness the full potential of this technology while mitigating security and privacy concerns. Scheduled for release early next year, the new platform, VMware Private AI Foundation with NVIDIA, will facilitate enterprise access to Nvidia's accelerated computing capabilities and generative AI software. This will enable the customization of models and the execution of generative AI applications, encompassing intelligent chatbots, virtual assistants, search engines, and summarization tools. Recognizing that enterprise data required for training and customizing generative AI models is dispersed across various cloud environments and data center infrastructures, Nvidia and VMware intend to provide secure and cost-effective access to this data.
Paul Turner, Vice President of Product Management, vSphere at VMware, emphasized the importance of safeguarding data assets, suggesting that companies may opt to bring more generative AI workloads closer to their data centers rather than migrating data to public cloud services. According to Turner, enterprises may initially deploy the platform in relatively small-scale environments, such as clusters comprising two to four GPU-based servers. However, as enterprises seek enhanced performance for their generative AI projects, the platform can leverage Nvidia's storage capabilities, NVLink, and NVSwitch, integrated with VMware vSphere, to support configurations of up to 16 GPUs. This enables an accelerated data feed into the AI models.
These pooled resources can be efficiently shared across different departments, ensuring cost-effectiveness while preserving privacy and security. Customers will have the flexibility to run AI services in proximity to their data sources. The platform will feature Nvidia NeMo, a cloud-native framework that is part of Nvidia AI Enterprise, the operating system of the Nvidia AI platform. NeMo encompasses customization tools, guardrail toolkits, data curation tools, and pre-trained models, enabling enterprises to build, adapt, and deploy generative AI models across diverse environments.
Justin Boitano, Vice President of Enterprise AI at Nvidia, highlighted the timeliness of enterprise customization of generative AI as this transformative technology continues to influence every sector. Boitano referenced research from consulting firm McKinsey, which estimates that generative AI has the potential to generate an additional $4.4 trillion in annual business value. This technology is poised to revolutionize industries by expediting ideation, automating complex tasks, enhancing creativity, and facilitating communication across various modalities.
Boitano concluded that generative AI would serve as indispensable assistants, seamlessly integrated into all facets of businesses across every industry, unlocking substantial untapped potential.