Dell Unveils Full-Stack Generative AI Solutions

Dell unveiled solutions today that it said would simplify the deployment and adoption of full-stack generative AI for its enterprise clients. The offerings build on a recently announced partnership with Nvidia called Project Helix.

"Customers, big and small, are using their own data and business context to train, fine-tune and inference on Dell infrastructure solutions to incorporate advanced AI into their core business processes effectively and efficiently," said Jeff Clarke, Dell co-COO, on a recent investor call.

First, Dell Generative AI Solutions will encompass Dell's IT infrastructure, PCs and services to provide tools to deliver solutions from desktops to data centers, edge locations and public clouds. The computer giant said this service will include Dell's Precision workstations, PowerEdge servers, PowerScale scale-out storage, ECS enterprise object storage, and services.

Dell Validated Design for Generative AI with Nvidia is an inferencing blueprint meant to accelerate the deployment of a modular, scalable and secure generative AI platform in the enterprise. Traditional inferencing methods struggled to scale and support large language models for real-time outputs and also ensure data can be easily used by AI infrastructure. Dell said this new solution will accelerate the process, generating "higher quality, faster time-to-value predictions and decisions with their own data."

Related:Nvidia partners with Microsoft, Dell on Generative AI Projects

This blueprint is pre-tested for use with Dell infrastructure with a choice of Nvidia Tensor Core GPUs, Nvidia AI Enterprise software, Nvidia NeMo end-to-end framework and with Dell software at its core. Clients can use it with scalable unstructured data storage; it is available now through Dell Apex and other channels for on-premises deployment with cloud consumption and management.

With a reference architecture for inferencing, Dell said it is "delivering a roadmap for how to cost efficiently take advantage of pre-trained models via Nvidia NeMo framework - rather than building and training your own model from scratch," wrote Varun Chhabra, Dell senior vice president of Infrastructure and Telecom Marketing, in a blog post. "Pre-trained models can also be further fine-tuned with smaller amounts of task-specific data, while giving access to a range of functionalities ready to use such as language translation, for example."

Dell Professional Services lets clients develop a generative AI strategy to identify optimal use cases and ways to achieve their goals, tap full-stack implementation and adoption services to apply the AI platform to specific use cases such as customer operations or content creation. Afterwards, Dell will help the enterprise scale through its managed services, training or resident experts.

Dell also said its Precision workstations will help AI practitioners develop and fine-tune generative AI models locally before deploying at scale. They carry up to four Nvidia RTX 6000 Ada Generation GPUs each that Dell said will "run AI software frameworks 80% faster than the previous generation." The built-in AI software for them is Dell Optimizer.

Dell said Precision workstations 7960 Tower, 7865 Tower and 5860 Tower with Nvidia RTX 6000 Ada Generation GPUs will be available worldwide in early August. Dell Optimizer will be released globally on Aug. 30 on select Precision mobile workstations.

The latest solutions build upon Dell's Project Helix, which is a collaboration with Nvidia to "make it easier for businesses to build and use generative AI models on-premises." Announced in May, Helix aimed to offer full-stack solutions with pre-built tools and technical expertise based on Dell and Nvidia infrastructure and software. The overarching goal of Helix is to offer optimized hardware and software solutions throughout the generative AI lifecycle: from infrastructure provisioning, modeling, training, fine-tuning, application development and deployment to deploying inference and streamlining results.