Dell is hoping to make hay while the Generative AI sun shines by cramming its portfolio with options for customizing enterprise AI models.

Dell’s focus on AI is no accident, with the company admitting its profits have been given a boost by all the recent corporate interest.

“Artificial intelligence is a strong tailwind for all things data and compute,” Dell vice chairman and chief operating officer Jeff Clarke said recently.

The Texas tech giant is also adding some extra professional services, and said it would work with data analytics outfit Starburst to access multiple data sources for AI models.

Dell rolled out its first Generative AI lineup at the end of July, but the focus there was mostly on validated designs for inferencing using existing models. But Dell wants to expand its lines to other stages in the AI pipeline and is now turning to model customization and tuning.

‘Hesitancy’ due to tech complexity and more

While there has been much interest in Generative AI ever since OpenAI unleashed ChatGPT onto an unsuspecting world, Dell claimed there is “a lot of hesitancy” among enterprise customers in adopting the technology due to technical complexity, security concerns, and data governance worries.

In addition, early adopters have found that the pre-trained models available may not be an exact fit for their requirements, and thus some tweaks inevitably need to be made.

“Due to the complexity, and the necessary customization that enterprises need to do with the data, many find that they need to rapidly expand beyond pre-trained models with their corporate data, creating requirements to customize large language models (LLMS),” said Carol Wilder, Dell VP for cross portfolio software and solutions.

To address this need, the company has announced the Dell Validated Design for Generative AI: Model Customization, which now supports both model tuning and inferencing, allowing users to more quickly deploy GenAI models, Dell claimed.

This will be available from late October through traditional channels and Dell’s APEX IT as-a-service arrangement. Like the existing validated design, it is a managed platform intended to run on Dell hardware such as the PowerEdge XE9680, which can be configured with up to eight Nvidia H100 GPUs, plus Dell storage, and likewise includes Nvidia’s AI Enterprise suite and NeMo framework for large language models (LLMs).

Likewise, the expanded professional services now include Data Preparation Services, intended to clean up customer data and ensure it is the correct format for AI projects; Implementation Services to aid companies get to an operational GenAI platform for inferencing and model customization; plus Education Services to provide workers with the necessary critical skills for GenAI.

“We can work with customers to establish a strategy, understand and define the vision and the solution architecture, and identify realistically the use cases and models in the business outcome that our customers are looking for,” Wilder claimed.

These additional professional services are set to be available in select countries from late October, Dell said.

Not strictly part of the Generative AI offerings, Dell disclosed it is working with Starburst on what it calls its vision of an “open, modern data lakehouse” in order to support a broader set of AI workloads as well as analytics.

According to Greg Findlen, Dell senior VP for Data Management, this is intended to be multi-cloud with an open, scale-out architecture and a unified data stack so that customers can run any workload on top.

“Customers can’t continue to have to choose between on premises and putting all of their data in the cloud. We need to help them reduce data movement so they can leverage the data gravity for where the data is and not have to consolidate all the data before they can use it,” Findlen said.

Many customers are trying to deal with this challenge by pulling together different pieces of technology by themselves, he claimed, so Dell’s goal is to build a more integrated set of capabilities that businesses can rely on.

“Customers feel like they’re on a treadmill where they need to consolidate all of their data in one place before their data scientists can start to use it. With this solution, customers can leverage the data where it exists,” Findlen claimed.

It is also intended to offer a simplified management experience with integrated enterprise grade support and security across the stack.

The first fruits of this project are set to be delivered in the first half of 2024. It will be based on the Starburst Enterprise Platform and run on Dell PowerEdge servers with Dell object and file storage.

AI accounted for 20 percent of server revenue in the first half of the year, the company disclosed along with its results for the second quarter of its financial year 2024. ®