Nvidia Buys Bright Computing for HPC Cluster Management

By | Managed Services News

Jan 13

The addition of Bright Cluster Manager fills gap in Nvidia’s DGX cloud data center infrastructure platform.

Nvidia has added high-performance computing (HPC) cluster management to its data center portfolio with its acquisition of Bright Computing. Nvidia announced and closed the transaction this week. It isn’t saying how much it paid for Bright Computing.

Founded in 2009 and based in Amsterdam, Bright Computing promises to fill an important gap in Nvidia’s DGX data center platform. Companies in financial services, health care, manufacturing and other industries that have formidable HPC implementations use the company’s Bright Cluster Manager.

More than 700 organizations use Bright Computing’s HPC software including Boeing, NASA, John Hopkins University and Siemens, according to Nvidia. Both Bright and Nvidia have worked together as partners for more than 10 years. Nvidia and its channel partners often integrate Bright’s software with Nvidia’s GPUs and its CUDA parallel computing platform.

Now Nvidia is expanding into the cloud and data center with DGX, systems designed to provide scalable AI infrastructure. Nvidia designed DGX to enable more organizations to model, build and run AI-based solutions, which increasingly require HPC infrastructure.

HPC Embedded Into DGX

Nvidia believes owning Bright Computing and incorporating its software will make it easier to build and sell HPC data centers. That’s because many organizations lack HPC expertise, said Charlie Boyle, VP and general manager of Nvidia’s DGX systems group.

Nvidia's Charlie Boyle“We’re now seeing a lot of enterprise customers wanting to use modern AI applications, whether it’s big language models, recommenders, those types of things,” Boyle told Channel Futures. “All of those systems require a large scale, clustered compute environment to execute those applications. One of the big challenges is that anyone other than a classical HPC user doesn’t really have the in-house experience to run these large clusters that are required for modern AI and other things.”

Bright Cluster Manager is designed to manage large HPC implementations consisting of x86, Arm, various GPUs and most recently DGX. Because Nvidia didn’t have its own cluster management offering, until now the company typically recommended Bright. But Boyle said despite Nvidia’s strong endorsement of Bright, many customers wanted HPC cluster management embedded in its solutions.

“Being a great partner and talking to them over the years, this just made a lot of sense,” Boyle said.

Bright Computing CEO Bill Wagner agreed.

“We’ve got a lot of joint customers, we have strong alignment in complementary products, and we share a common view of the future, which obviously heavily influences product direction,” Wagner said.

Expanding Channel Reach

Almost all of Bright’s channel partners also have partnerships with Nvidia, Wagner added.

Bright Computing's Bill Wagner

Bright Computing’s Bill Wagner

“It doesn’t change any of the reseller relationships that we have,” he said. “In fact, it just opens up an entirely new set of resale relationships for us that are in the Nvidia partner network that we currently don’t have relationships with.”

Nvidia’s Boyle underscored that while most of Bright’s partners are also Nvidia partners, that only accounts for a subset of the potential reach.

“Obviously, we have a lot more partners in our network than Bright had in theirs, so that’s going to open up new opportunities for them,” Boyle said. “But we’re also not changing anything in their current routes to market because they were the same route to market that we use.”

Bright will help Nvidia offer more complete enterprise data center platforms, noted Karl Freund, founder and principal analyst of Cambrian-AI Research.

“Strategically, Nvidia has been checking off all the boxes needed to be considered a full-range supplier of computing solutions for cloud and the enterprise, including servers, networking, and software solution frameworks,” Freund noted in a post on Forbes. “Now they can add cluster management to the list.”

 

About the Author

>