Nvidia wants to make it easier and less expensive for remote workers to collaborate on graphics-intensive workloads through a newly revamped program that puts the chipmaker's highest performing GPUs in the hands of cloud service providers.

With the onset of COVID-19, "a lot of engineers went home and had to leave their multi-GPU workstations at their desk," said Anne Hecht, senior director of product marketing for GPU virtualization software at NVIDIA. "We have had customers contact us wanting to know 'how do I get my engineers up and running from home? Laptops aren't sufficient for their workloads.'"

Nvidia sees the pandemic shift to remote work — a trend that an IDC report from June found was likely to be permanent for up to 52% of workers — as an opportunity for cloud service providers to fill the void.

Cloud-based GPUs are nothing new for Nvidia. In fact, Nvidia's Cloud Service Provider program is nearly three years old. But, with the onset of the pandemic, Nvidia reinvested in the program, which has grown more than 60% since the start of the year.

This month, Cloudalize, CoreWeave, Dizzion, E2E, IronOrbit, and Paperspace joined 32 other service providers added this year alone.

According to Hecht, running graphics-intensive workloads in the cloud has numerous advantages over traditional desktop workstations.

One of the biggest is the ability to share resources between multiple users, Hecht explained. "Depending on how much GPU acceleration they need, you might be able to put four users on a single GPU."

And while Hecht admits that a single user won't get the performance of a dedicated GPU, many professionals no longer have access to their workstations.

On the flipside, remote workers can also access larger banks of GPUs than would have been possible in a single workstation. And since cloud resources are billed based on the resources used, it is often far more economical, she adds.

Hecht adds that unlike the major cloud-providers that have deployed machine learning and artificial intelligence-focused GPUs, like Nvidia's V100 and A100 chips, many service providers are opting for the RTX 6000 and RTX 8000, which are designed specifically for workloads like rendering and computer-aided design.

Beyond flexibility, Hecht argues cloud-based graphics also enable faster and more secure collaboration by allowing remote workers to work together without needing to download massive data sets and then re-upload them for review.

"It's a lot easier to do that collaboration if the model and all the data is stored in a centralized data center, whether that data center is in the cloud or in an on-prem environment," Hecht said.