Nvidia Unveils AI Partnerships with US Government and Equinix

Today, Nvidia unveiled two key new partnerships — one with the US Government’s National Science Foundation and one with the world’s largest data center colocation provider, Equinix.

At a high level, the partnerships mark an attempt to increase the accessibility of artificial intelligence (AI) to researchers and scientists in the public and private sectors.

Under the partnership with the US National Science Foundation, Nvidia will be providing $30 million of compute infrastructure, including $24 million of Nvidia DGX Cloud Compute, to support the establishment of the government’s National Artificial Intelligence Research (NAIRR) pilot program.

Nvidia’s collaboration with Equinix will give enterprise access to a fully managed Nvidia AI supercomputing service, which will be installed and operated by Equinix on each customer’s privately-owned Nvidia infrastructure, with key services deployed to Internet Business Exchange (IBX) data centers globally.

Table of Contents Table of Contents

1. Key Takeaways
2. Nvidia-US Government Partnership: Making AI Research Accessible
3. Nvidia's-Equinix Partnership: Making AI Accessible to the Enterprise
4. The Bottom Line

Show Full Guide

Table of Contents

1. Key Takeaways
2. Nvidia-US Government Partnership: Making AI Research Accessible
3. Nvidia's-Equinix Partnership: Making AI Accessible to the Enterprise
Show Full Guide
4. The Bottom Line

The news comes just months after Executive Order 14110, which called for the launch of the National Artificial Intelligence Research Resource pilot program to encourage safe and responsible AI development.

We speak with Nvidia about what these partnerships mean, why they matter, and offer our own views on their significance.

Nvidia-US Government Partnership: Making AI Research Accessible

In broad terms, Nvidia’s partnership with the US government has the potential to democratize AI development by giving American researchers a mechanism to access compute resources at selected locations, including the Oak Ridge National Laboratory and the National Center for Supercomputing Applications.

Arati Prabhakar, assistant to the President for Science and Technology and director of the White House Office of Science and Technology, said:

“Today’s announcement makes progress on President Biden’s goal to advance Responsible AI so that everyone in America can benefit from this powerful technology.

 

“The National AI Research Resource pilot will give researchers access to critical data and compute, catalyzing action to achieve America’s great aspirations.”

Researchers can apply for access to Nvidia’s compute infrastructure under the program by creating a proposal and applying via an online portal on the official website.

Proposals are open until March 1, 2024, after which there will be another call for proposals sometime in Spring 2024.

The types of projects the pilot is looking for include — but are not limited to — testing, evaluating, verifying, and validating AI systems, improving model performance, increasing the interpretability or privacy of models, reducing model vulnerability, or making sure models align with societal values and obey safety guarantees.

Nvidia’s-Equinix Partnership: Making AI Accessible to the Enterprise

On the other side of the fence is Nvidia’s Equinix partnership, notable because it provides an opportunity for the organization to bring its services to the world’s largest colocation provider, with a reach of 250 data centers across six continents, over 10,000 customers, and over 260 Fortune 500 companies.

The new service, Equinix Private AI with Nvidia DGX, provides a solution for enterprises that want to own their own AI infrastructure while leveraging Equinix’s skills at managing and deploying AI-driven operations.

Charlie Boyle, vice president of DGX Systems at Nvidia, told Techopedia:

“Generative AI has increased the strategic importance of AI, making it even more important for enterprises to have access to powerful AI infrastructure.

 

“This new service allows organizations to reduce risk when adopting or scaling their AI capabilities, while retaining the benefits of owning the infrastructure.

 

“It lets them deploy faster and outsource operations so they can focus on innovating rather than deploying and managing infrastructure, raising the quality of the AI-powered products and services they create.”

Boyle said that Equnix Private AI with Nvidia DGX is a fully managed service that will enable enterprises to acquire and manage their own Nvidia DGX AI supercomputing infrastructure for building and running custom generative AI models.

The service will also provide access to data center infrastructure like Nvidia DGX SuperPOD?and in Nvidia Nemo, a framework architecture for training and running LLMs.

Fundamentally, this partnership helps to provide the infrastructure necessary to deploy AI in the enterprise.

The Bottom Line

Nvidia’s latest partnerships with the US National Science Foundation and Equinix help to solidify the vendor as a key enabler of the move toward responsible AI development.

In terms of the Equinix partnership, there’s a cost in adopting AI, and generative AI in particular, due to the prohibitive costs of training and running large language models (LLMs). By some estimates, models like GPT-3 could cost over $4 million to train.

In this offering, infrastructure from Nvidia can help make AI more cost-efficient for enterprise customers, as few organizations have the internal resources to build and run models in-house.

Increasing access to compute infrastructure will ultimately be the tide that raises all boats in the AI research race.