High-performance computing is a rapidly growing market, set to grow by about $14 billion from 2022 to 2027, with expected developments fueling continued expansion.

With cloud scale and semi-conductor silicon innovations and a growing cross-disciplinary focus and new technical advancements, we’ve entered a new generation for HPC with the introduction of data, machine learning and generative artificial intelligence. Vultr, a registered trademark of The Constant Company LLC, is the world’s largest privately held cloud compute platform. It uses powerful cloud GPUs, including the Nvidia H100, and around 32 global data centers to deliver HPC acceleration to assist with generative AI, data analytics and model training.

“What we think is very exciting is what’s to come in 2024, because it’s time to move beyond experimentation and single large-scale training clusters and moving beyond the art of the possible to the realization of actual outcomes,” said Kevin Cochrane (pictured), chief marketing officer at Vultr. “This is where it gets exciting, because this is where enterprises are going to start architecting for the future.”

Cochrane spoke with theCUBE industry analysts John Furrier and Lisa Martin at SC23, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed using HPC to power AI and machine learning, the ever-evolving edge and cross-disciplinary integration from silicon to cloud. (* Disclosure below.)

Optimizing deployment and security

Companies across the world and industries are looking to take advantage of the powerful potential of artificial intelligence, and leaders need to know how to transition from training AI models to using the model for real-world tasks, a process known as inference, Cochrane explained.

“We not only need to stand up training clusters, but we also need to actually optimize how those trained models are then deployed and then fine-tuned in the region against localized data sets,” he said. “Once fine-tuned … they can then be deployed efficiently to large-scale inference clusters running the new GH200 superchip from Nvidia, powered by Vultr, and deliver predictions for incoming user requests.”

When it comes to what Vultr is doing differently, one big feature that sets the company apart from the competition is being the only independent cloud provider that can provision Nvidia GPUs on Dell hardware across the globe, at the same scale as Amazon, according to Cochrane. Vultr is also a secure and compliant independent cloud, keeping application, security, compliance and governance in mind with their clients.

“We have a unique set of services to enable the entire pipeline of distribution of both your application artifacts, your code, co-resident with your generative AI artifacts, your models,” Cochrane said. “We’re the only vendor that basically provisions Kubernetes clusters for your application models, your application code so that you can actually truly reinvent your enterprise.”

Cochrane also shared some personal insights into what outcomes will start to emerge as we enter 2024. One outcome he hopes is achieved is the establishment of a policy framework that all enterprises can adhere to as the first AI-powered applications are being built.

“That way we’re leveraging National Institute of Standards and Technology standards for things like explainability and accuracy so we don’t wind up having unintended consequences when we’re trying to do good,” Cochrane said. “There are standards out there with NIST. I encourage everyone to read those standards and be thoughtful about what you’re doing.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of SC23:

(* Disclosure: TheCUBE is a paid media partner for SC23. Neither Dell Technologies Inc., the main sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE