Jensen Huang's CES 2025 address: Key takeaways from NVIDIA CEO’s vision for AI computing

Note: This feature was first published on 7 January 2025.

At CES 2025, NVIDIA’s CEO and founder, Jensen Huang, took the stage to lay out his vision for the future of AI (artificial intelligence) computing – giving attendees and viewers online a glimpse of how the next generation of technology will unfold, while powered by NVIDIA’s formidable hardware and software. From the announcement of the GeForce RTX 50 series to advancements in autonomous vehicles and – surprisingly – robotics, here's a look at the most significant moments from Jensen’s CES address.

The keynote kicked off with a nod to the ongoing transformation that AI is driving across nearly every industry. In a clear statement of intent, Huang spoke about the profound ways in which AI is now embedded into the fabric of modern computing. It’s not just about more efficient machines; it’s about creating entirely new possibilities for how we interact with technology, from gaming to autonomous vehicles to everyday computing tasks. This is a paradigm shift that NVIDIA, with its powerful GPUs and AI platforms, is uniquely positioned to lead.

A standout moment from Huang's keynote was the announcement of the GeForce RTX 50 series GPUs. While NVIDIA has been at the cutting edge of graphical performance in the consumer and gaming segments for years, the new RTX 50 series represents a significant leap forward, not just for gaming but for AI-driven applications. Jensen emphasised that these new GPUs are designed to meet the increasing demands of AI, enabling a new era of real-time, AI-assisted graphics rendering. The new cards will offer enhanced ray tracing capabilities, better performance in AI workloads, and a more immersive gaming experience overall. The RTX 50 series will not just be for gamers but for anyone working with AI-powered tools, whether in content creation, data analysis, or simulation.

But while the RTX 50 series certainly stole the spotlight in terms of gaming and graphics, the keynote was far more than just a gaming-focused announcement. Jensen also spent a considerable amount of time discussing the future of autonomous vehicles. It’s a space that NVIDIA has been investing heavily in for years, and Jensen is convinced that autonomous vehicles will become the first multi-trillion-dollar robotics industry. He laid out the case for why self-driving cars are on the verge of mass adoption, predicting that we are about to witness an acceleration in their development and deployment. It’s a bold claim, but Jensen notes that there are already over a billion cars on the road worldwide, collectively driving a trillion miles annually. The need for highly autonomous or fully autonomous vehicles will only continue to grow as this industry scales.

To power the next generation of autonomous vehicles, Huang introduced Thor, the next-gen processor designed specifically for the AI and sensor-heavy demands of self-driving technology. According to Jensen, Thor boasts 20 times the processing power of its predecessor, Orin, which is the current standard for autonomous vehicles. The processor is capable of handling vast amounts of data from sensors like cameras, LiDAR, and radar, processing it in real-time to make decisions about vehicle navigation. Jensen described Thor as a "robotics computer" capable of much more than just driving cars. It’s a versatile processor that could also be used in a variety of other robotics applications, from industrial robots to humanoid machines.

Jensen also made a point of highlighting the importance of safety in autonomous driving during his keynote. To that end, NVIDIA has worked on its Drive OS software, which has now been certified to the highest automotive safety standard: ISO 26262, an important milestone that sets the company apart in the race for self-driving technology. This certification can also be seen as a vote of confidence by regulators of NVIDIA’s commitment to ensuring that its autonomous driving systems are also fully compliant with the most stringent safety requirements in the industry.

Jensen didn’t stop there. His keynote also delved into the broader implications of AI for robotics. He teased what he described as the “ChatGPT moment for general robotics,” signalling that the breakthroughs in AI he’s seeing today are laying the groundwork for rapid advancements in robotics, too. According to Jensen, general-purpose robots capable of operating in human environments – whether it’s in offices, homes, or factories – are on the cusp of becoming a reality. The key to achieving this, he noted, is the ability to train these robots effectively. While training data for self-driving cars is relatively easy to come by – simply collect data from cars on the road – training robots for complex, human-centric tasks is much harder. To address this, NVIDIA is turning to synthetic data generation, using its Omniverse and Cosmos platforms to create realistic virtual environments that robots can be trained in before they ever interact with the real world.

NVIDIA’s Isaac Group is leading the charge in this area, and Jensen provided a detailed breakdown of how its platforms are accelerating the development of general-purpose robots. Isaac GR00T, for instance, allows developers to generate vast amounts of synthetic data by simulating human movements. A handful of human demonstrations can be used to create millions of variations, which are then used to teach AI systems how to perform complex tasks. The key here is the ability to train robots in virtual environments, where developers can gather all the data they need without worrying about the risks of physical damage or wear and tear. Huang is confident this approach could massively speed up the development of humanoid robots, which could have applications in areas as diverse as healthcare, manufacturing, and service industries.

He also gave a sneak peek into the future with Project DIGITS, an ambitious initiative aimed at creating the world’s smallest AI supercomputer. This supercomputer is designed to be cloud-connected, allowing developers to tap into powerful AI computing resources from their desks. It’s part of NVIDIA’s larger push to democratise access to advanced AI tools and supercomputing. Unlike traditional supercomputers, which require massive data centres and specialist infrastructure, Project DIGITS promises to make AI computing more accessible to smaller companies, startups, and individual developers. With the help of NVIDIA’s Grace Blackwell chip – developed in collaboration with MediaTek – this small, portable AI workstation could drastically change not only how we use, but also how we think about AI hardware.

Finally, Jensen ended his keynote by reiterating the pervasive nature of AI. He believes that every engineer, every artist, and every creative person will soon need to incorporate AI into their daily workflows. The age of AI-powered computing is here, and NVIDIA intends to make it as accessible and impactful as possible. Whether it’s through the GeForce RTX 50 series GPUs that power immersive gaming experiences, the AI chips that drive self-driving cars, or the robots that will disrupt industries, NVIDIA is shaping the future of technology in ways that will have far-reaching implications for industries and consumers alike.

Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.