Dell Technologies, in partnership with VMware, launched two products that aim to make it easier for customers to run artificial intelligence (AI) workloads in VMware environments. The company calls these Dell EMC Ready Solutions, and they combine Dell EMC hardware systems with VMware’s Cloud Foundation software stack using virtualization technology that VMware acquired when it bought Bitfusion last year.

Bitfusion’s technology enables the virtualization of hardware accelerators like GPUs, field-programmable gate arrays (FPGAs), and ASICs to bolster computing power. Its software platform decouples those specific physical resources from the servers where they reside so that those resources can be shared across the IT environment. The virtualization of these accelerators is traditionally done in hardware, which limits how those resources can be allocated across an organization.

When VMware bought Bitfusion last summer, it said the move would boost its ability to support AI and machine learning workloads in its core vSphere virtualization platform. Since then VMware integrated Bitfusion into vSphere, and the new Dell EMC Ready Solutions use the Bitfusion feature in vSphere 7 to manage and allocate a pool of virtual GPU accelerators. This means that customers don’t need to buy individual servers for AI workloads and instead can share these resources across data centers.

Perfect Storm: Too Much Data, Not Enough AI Adoption

The amount of data generated globally is expected to explode to 175 zettabytes by 2025, up from 33ZB in 2018, said Ravi Pendekanti, SVP of server and infrastructure systems at Dell Technologies. “Over 90% of [enterprise] applications are expected to use AI in the next five years,” he continued, adding that only 14.6% of companies report widespread deployment of AI capabilities, according to a Forbes study from January.

However, resources like GPUs that could speed up AI adoption are typically trapped in individual servers and often underutilized — Dell puts this figure at less than 15% of GPU capacity.

“So here is a perfect storm: lots of data being created,” Pendekanti said. “And a lot of applications that intend to use artificial intelligence, but less than 15% are in a mode to actually utilize this.”

Dell’s plans to help speed enterprise AI adoption centers on virtualizing IT resources.

The first of the two new products, Dell EMC Ready Solutions: GPU-as-a-Service, uses vSphere 7 with Bitfusion to create these virtual GPU pools for AI developers and data scientists to use and share. The GPU-as-a-Service also uses vSphere 7 support for Kubernetes and containerized applications to run AI workloads across hybrid cloud environments.

The second product, Dell EMC Ready Solutions for Virtualized HPC (vHPC) targets high-performance compute (HPC) workloads and uses vSphere 7 with Bitfusion to virtualize HPC and AI operations. According to Forrester, Dell EMC Ready Solutions for vHPC can enable up to 18 times faster AI model development, and up to 20% faster hardware configuration and integration compared to self-installation. Dell estimates this will provide a return on investment of up to 111%.

Both of these are globally available now. Additionally, factory installation of VMware vSphere with Bitfusion will be available on Dell EMC PowerEdge servers in July.