Nvidia AI Foundry
For enterprises to adopt AI, it needs to become much more accessible and affordable. Nvidia has relaunched its AI Foundry to help businesses integrate and customize AI solutions tailored to their needs — without starting from scratch or making overwhelming financial investments. Nearly a year ago, Nvidia also introduced AI Foundry Services to support these initiatives. But, this year, they relaunch it in July 2024.
The timing is ideal, as many companies are moving toward AI and need a comprehensive ecosystem that simplifies their AI journey without requiring a heavy financial commitment. The difference between the launch this year versus last year is the incorporation of the NIMs (Nvidia Inferencing Microservices) and more public cloud beyond Azure.
What is Nvidia AI Foundry ? The Nvidia AI Foundry is a combination of software, models, and expert services to help enterprises start and complete their AI journey. Similar to how TSMC manufactures chips designed by other companies, NVIDIA AI Foundry supplies the infrastructure and tools for businesses to develop and customize their own AI models — leveraging DGX Cloud, foundation models, NVIDIA NeMo software, NVIDIA’s expertise, and a robust ecosystem of tools and support. The result is NVIDIA NIM™ — an inference microservice containing the custom model, optimized engines, and a standard API — that can be deployed anywhere. Here is the pictorial diagram that explain the whole process. Notice it support on public cloud and Oracle Cloud Infrastructure only.
How does this differ from using Retrieval-Augmented Generation (RAG) with a large language model? RAG is effective for enhancing an LLM with company-specific data. However, NVIDIA claims that the AI Foundry can create a fully customized model that is up to ten percentage points more accurate than a basic RAG approach. That increase in accuracy can be the difference between a highly effective model and one that might ultimately be discarded.
Accenture has leveraged the NVIDIA AI Foundry to enhance its own internal enterprise functions and has used these insights to develop the Accenture AI Refinery, aimed at helping clients achieve similar results. Deloitte is pursuing a similar approach.
As enterprises strive to adapt and implement custom AI models for their specific needs, this solution offers a streamlined path to becoming an AI-driven organization.
