On Tuesday, Nvidia made a low-key but impactful announcement: the launch of its AI
Foundry service. This new offering is set to help businesses craft and deploy custom
large language models tailored to their specific needs. It’s a strategic move by Nvidia to
grab a bigger slice of the rapidly expanding enterprise AI market.
So, what’s AI Foundry all about? It’s a blend of Nvidia’s advanced hardware, software
tools, and deep expertise, designed to let companies create bespoke versions of
popular open-source models, like Meta’s fresh Llama 3.1. This service hits the market
as businesses are eager to harness the power of generative AI while keeping a tight
grip on their data and applications.
Customization Meets Accuracy: Nvidia’s Game-Changer
“This is the moment we’ve been waiting for,” Kari Briski, Nvidia’s VP of AI Software, told
VentureBeat. “Enterprises rushed to understand generative AI, but another key
development was the availability of open models.”
Nvidia’s AI Foundry aims to make the process of tailoring these open models for
specific business needs simpler and more efficient. According to Briski, the company
has observed nearly a ten-point increase in accuracy just by customizing these models.
With AI Foundry, companies gain access to a treasure trove of pre-trained models, top-
notch computing power via Nvidia’s DGX Cloud, and the NeMo toolkit for fine-tuning
and assessing models. Plus, Nvidia’s AI experts are on hand to guide the process.
“We provide the infrastructure and tools for companies to develop and customize AI
models,” Briski explained. “Enterprises bring their own data, and we have DGX Cloud,
which partners with many cloud providers.”
Nvidia’s NIM: The Future of AI Model Deployment?
Alongside AI Foundry, Nvidia unveiled NIM (Nvidia Inference Microservices), a tool that
packages customized models into containerized, API-friendly formats for easy
deployment. This is a major milestone for Nvidia. “NIM is a customized model in a
container accessed by a standard API,” Briski said. “It’s the result of years of work and
research.”
Industry experts view this as Nvidia’s strategic move to expand beyond its core GPU
business, aiming to become a comprehensive AI solutions provider rather than just a
hardware maker.
The Timing Is Everything: Nvidia’s Strategic Move
The timing of Nvidia’s announcement is intriguing, coinciding with Meta’s Llama 3.1
release and growing concerns about AI safety and governance. By offering a service
that lets companies create and control their own AI models, Nvidia could be tapping into
a market of businesses that want the advantages of advanced AI without the risks
associated with using generic public models.
But, there are some uncertainties about the long-term impact of widespread custom AI
model deployment. Potential issues include fragmentation of AI capabilities across
different sectors and the challenge of maintaining consistent standards for AI safety and
ethics.
As competition heats up in the AI space, Nvidia’s AI Foundry is a bold bet on the future
of enterprise AI. The success of this initiative will depend largely on how well
businesses can use these custom models to drive real-world innovation and value in
their industries.
Join our daily and weekly newsletters for the latest updates and exclusive content on industry- leading AI coverage.
