Nvidia releases software and services to drive rapid adoption of GenAI
The software and services are included in the Nvidia AI Enterprise product, which costs $4,500 per graphics processor used per year | Photo: Bloomberg
By Ian King
Nvidia Corp., the world’s most valuable chipmaker, has announced a series of updates to its software offerings that should make it easier for a wider range of companies to use generative artificial intelligence.
The chip company officially released what it calls NIMs, or Nvidia inference micro services, software packages that solve many of the logistical problems required to deploy AI for a specific purpose. Generative AI, the technology behind chatbots, speech recognition and other automated human interactions with computers, often requires the orchestration of countless pieces of hardware, software and information retrieval. Many companies don’t have the expertise, so Nvidia is trying to do it for them, for a fee.
Nvidia founder and CEO Jensen Huang unveiled his company’s latest offerings Monday at the Siggraph conference in Denver. He’s at the graphics technology conference — the latest in a series of public event appearances — trying to convince as many industries as possible to adopt his technology and make AI ubiquitous across the economy. He’ll later appear onstage at the conference with Meta Platforms Inc. CEO Mark Zuckerberg.
Nvidia’s chips have become central to the surge in building new systems that support artificial intelligence. The company’s revenue doubled last year and is on track to double again in the current fiscal year.
The software and services are included in the Nvidia AI Enterprise product, which costs $4,500 per graphics processor used, per year. The software and services are designed to run on Nvidia hardware.
“The Nvidia NIM is a comprehensive solution for implementing generative AI, simplified for developers but built for large-scale applications,” said Kari Briski, vice president of product management for AI and HPC software development kits at Nvidia.
Nvidia had made about 100 of its inference microservices available in preview and is now releasing finished versions. For example, Getty Images Holdings’ services are getting improved generation of high-resolution images by making the software better at understanding text prompts. Shutterstock Inc.’s Edify 3D image generator is going live with the ability to react to text or images.
According to Nvidia, most AI is used by knowledge workers to help with digital tasks. To make generative AI more widespread, Nvidia is providing software and services that let users of devices like Apple Inc.’s Vision Pro headset create virtual worlds. So-called virtual twins are used for tasks like training robotic computers to act more like humans, freeing developers from having to do it manually, Nvidia says.
First print: Jul 30, 2024 | 08:21 AM IST