At Nvidia’s CES 2025 keynote, CEO Jensen Huang wasted no time capturing the audience’s attention by unveiling the highly anticipated GeForce RTX 50 Series. Huang walked onstage holding the new graphics card, greeted by enthusiastic applause. While this was the highlight of the event, Nvidia had a host of other groundbreaking announcements to share at the Las Vegas tech conference. The company, pivotal in advancing generative AI, is now developing its own AI models, supporting robotics and autonomous vehicle innovation, and delivering cutting-edge computing tools to the public. Here’s a breakdown of everything announced during the keynote.
GeForce RTX 50 Series
The centerpiece of Nvidia’s announcements was the new GeForce RTX 50 Series GPUs, powered by the RTX Blackwell architecture. The lineup includes the flagship GeForce RTX 5090, along with the RTX 5080, 5070 Ti, and 5070. Boasting 92 billion transistors, these GPUs deliver a staggering 3,352 trillion AI operations per second (TOPS) and a memory bandwidth of 1.8TB/s. With specs that Huang described as “just a beast,” the RTX 50 Series represents a significant leap in graphics performance. Full details on specifications, pricing, and availability are covered by Mashable’s Chance Townsend and Alex Perry.
Cosmos World Foundation Models
Nvidia is venturing into robotics with the introduction of Cosmos World Foundation Models (Cosmos WFM). These models, designed for training robotics systems, are now available as an open-license platform on GitHub. By offering Cosmos WFM to the wider developer community, Nvidia aims to democratize access to advanced robotics tools and expertise. Huang hinted at a transformative moment for robotics, stating, “The ChatGPT moment for general robotics is just around the corner.”
AI Foundation Models for RTX AI PCs and Llama Nemotron
Nvidia also unveiled AI foundation models tailored for LLM (large language model) development on RTX PCs. Delivered as Nvidia NIM microservices, these models leverage the GeForce RTX 50 Series GPUs. Leading PC manufacturers will soon launch systems compatible with NIM, signaling the arrival of AI PCs for consumer use.
Additionally, Nvidia introduced the Llama Nemotron family of LLMs, built on Meta’s open-source Llama models. These models excel at tasks like instruction following, chat, function calling, coding, and math, while being optimized for Nvidia-accelerated computing environments. The Llama 3.1 Nemotron 70B is now part of Nvidia’s API catalog.
Project Digits
Continuing its mission to empower developers, Nvidia announced Project Digits, a compact supercomputer resembling a Mac mini. Powered by the GB10 Grace Blackwell Superchip, Project Digits can run LLMs with up to 200 billion parameters entirely on-premises, eliminating the need for cloud infrastructure. Priced at $3,000, this desktop AI supercomputer is an accessible solution for small businesses and solo developers. The device is expected to launch in May.
DRIVE Hyperion AV Platform
In the realm of autonomous vehicles, Nvidia introduced the DRIVE Hyperion AV platform, powered by the AGX Thor system-on-a-chip (SoC). This comprehensive platform includes sensors, safety systems, and the DriveOS operating system, providing automakers with an end-to-end solution for building autonomous vehicles. Nvidia revealed that Toyota has joined its roster of partners, which already includes Mercedes-Benz, Jaguar Land Rover, and Volvo.
These announcements underscore Nvidia’s expanding role across AI, robotics, autonomous vehicles, and high-performance computing, setting the stage for a transformative year ahead.
