As the AI world races toward next-generation breakthroughs, Nvidia (NVDA+3.84%) fortified its position with a flood of new chips, software and services designed to keep the industry plugged into its expanding tech ecosystem.
Nvidia chief executive Jensen Huang announced a suite of AI tools and updates during the first keynote speech at the Consumer Electronics Show (CES) on Monday, focusing on AI agents — systems that can complete tasks autonomously. Huang demonstrated these capabilities through animated versions of himself in different outfits, showing how AI agents could handle roles like customer service, coding, and research assistance.
“The IT department of every company is going to be the HR department of AI agents in the future,” Huang said in the keynote.
The chipmaker unveiled AI Blueprints, which will help companies build and deploy these AI agents using technology built on Meta’s (META+3.38%) Llama models. These “knowledge robots,” as Nvidia describes them, can analyze large amounts of data, summarize information from videos and PDFs, and take actions based on what they learn. To make this happen, Nvidia partnered with five leading AI companies — CrewAI, Daily, LangChain, LlamaIndex, and Weights & Biases — who will help integrate Nvidia’s technology into usable tools for businesses.
“These AI agents act like ‘knowledge robots’ that can reason, plan and take action to quickly analyze large quantities of data, summarize and distill real-time insights from video, PDF and other images,” Justin Boitano, Nvidia’s vice president of enterprise AI software products said in a statement.
Nvidia also made a major push into robotics, unveiling new tools to help companies simulate and deploy robot workforces. The centerpiece is “Mega,” a new Omniverse Blueprint that lets companies develop, test, and optimize robot fleets in virtual environments before deploying them in real warehouses and factories.
“The ChatGPT moment for general robotics is just around the corner,” Huang said during his keynote. To back this claim, Nvidia announced a collection of robot foundation models, including new capabilities for generating synthetic motion data for training humanoid robots. The company says these pre-trained models were developed using massive amounts of data, including millions of hours of autonomous driving and drone footage.
A key part of this effort is the new Isaac GR00T Blueprint, which helps solve a major challenge in robotics: generating the massive amounts of motion data needed to train humanoid robots. Instead of the expensive and time-consuming process of collecting real-world data, developers can use GR00T to generate large synthetic datasets from just a small number of human motions.
“Collecting these extensive, high-quality datasets in the real world is tedious, time-consuming, and often prohibitively expensive,” Nvidia said in a press release. With the Isaac GR00T blueprint, developers can generate the large data sets needed to train AI models from a small number of human motions.
The new tools are already attracting major industry players. KION Group, a supply chain solutions company, is working with Accenture (ACN-0.91%) to use Nvidia’s Mega blueprint to optimize warehouse operations that involve both human workers and robots.
Nvidia’s efforts into the physical world didn’t stop there: it announced the expansion of its automobile partners, including with the world’s largest automaker, which will use its accelerated computing and AI for driving assistance capabilities and autonomous vehicles.
Toyota (TM-2.56%) will use Nvidia’s DRIVE AGX Orin system-on-a-chip (SoC) technology for its next-generation vehicles for driving assistance. Meanwhile, the chipmaker said Aurora (AUR-5.21%) and Continental will both use Nvidia’s DRIVE accelerated compute to deploy driverless trucks through a long-term strategic partnership.
On the consumer front — it is CES, after all — Huang showed off the next generation of its RTX Blackwell GPUs, with the four versions priced between $549 and $1,999. The popular gaming chips will be released later this month and in February.
And to cap off the keynote, Huang lifted the curtain on Project DIGITS, a palm-sized personal AI supercomputer powerful enough to run AI models with up to 200 billion parameters – for comparison, GPT-3 has 175 billion parameters while GPT-4 is rumored to have 1.76 trillion parameters. The device, which will be available starting at $3,000 from Nvidia and its partners in May, aims to put AI development in more hands.
“AI will be mainstream in every application for every industry,” Huang said in a statement. “With Project DIGITS, the Grace Blackwell Superchip comes to millions of developers. Placing an AI supercomputer on the desks of every data scientist, AI researcher and student empowers them to engage and shape the age of AI.”
Nvidia stock was up 4% before Huang’s CES keynote, with Bank of America (BAC+1.36%) reiterated its “buy” rating on Nvidia in a note on Monday, with an analyst describing CES as a “positive catalyst” for the company. The stock closed at a record high of $149.43.
The post Nvidia goes all in on AI agents and humanoid robots at CES appeared first on Quartz.
