What did we learn from Nvidia at CES 2026, where the AI giant pushed beyond every limit. Literally. And if that sounds exaggerated, wait until you hear the highlights.
In just one day, Nvidia announced developments touching every corner of technology, from data centers and AI factories to robotics, autonomous driving, gaming and of course new RTX graphics cards. Today though, we focus on what truly matters. The announcements that clearly show where the future is heading.
THE NEW STRATEGY
Nvidia presented the philosophy of the “AI Factory”. A complete AI infrastructure stack, from processing with Rubin, to security and storage with BlueField, all the way to the software layer with open models and robotics support. The goal is clear. AI must stop being a custom solution for every company and become repeatable infrastructure. Like the factories of the 20th century, but built for Artificial Intelligence.
This vision appeared everywhere. In the new deskside supercomputers, DGX Spark and DGX Station, capable of running models with up to one trillion parameters directly from your desk. In reference architectures that give enterprises ready made templates to build AI systems. In robotics solutions with Cosmos and Isaac Lab for physical intelligence. And even in gaming, with DLSS 4.5, the new Dynamic Multi Frame Generation, G-SYNC Pulsar monitors, and the expansion of GeForce NOW to Linux and Amazon Fire TV.
The common thread is Nvidia’s effort to unify everything under a single, coherent solution. Because AI is no longer a collection of isolated applications. It is an entire ecosystem.
ALPAMAYO
Now let’s move to the announcement that truly left everyone speechless and will affect our everyday lives.
Nvidia introduced Alpamayo, the first vision language action model for autonomous vehicles that actually thinks. It does not simply follow instructions. It uses reasoning, adapts, and can respond to situations it has never encountered before.
Partnerships have already been announced with Mercedes-Benz, Lucid, Uber and JLR. Mercedes will begin in 2026 with the new CLA and the MB.OS platform. These vehicles will feature the full DRIVE AV solution, with the goal of seeing Level 4 robotaxis on the road by 2027.
Alpamayo was trained on the Cosmos foundation model, which has driven millions of kilometers in simulation before ever touching real roads. The use of synthetic data, combined with open training material and datasets, provides transparency and safety in the model’s decision making.
This is not theory. Journalists and analysts took part in test drives in San Francisco with the 2026 CLA. Ninety percent of the route was driven autonomously, with only one intervention during a complex traffic jam involving two buses and a Waymo vehicle. The feeling was complete comfort.
Nvidia states that by 2028, full autonomous point to point driving for private cars will be achievable. You will simply tell your car where you want to go and it will take you there without any intervention.
RUBIN
At the same time, Rubin entered full production. This is Nvidia’s next major leap in AI infrastructure. It is the natural evolution of Grace Blackwell and integrates six new processors on a single board: CPU, GPU, NVLink switch, BlueField DPU, ConnectX NIC and Spectrum Ethernet switch.
The result is a system capable of delivering up to five times the AI inference performance and ten times the token generation efficiency. And it achieves this with only 1.6 times more transistors, thanks to exceptional co design across components.
Nvidia’s goal is to bring AI infrastructure to every level, from the data center all the way to the developer’s desk. Companies such as Amazon, Google, Microsoft, Meta, OpenAI, Dell, Cisco, Oracle, Anthropic and many others have already committed to adopting Rubin.
WHAT BANK OF AMERICA SAYS
Let’s also look at Wall Street’s perspective. According to Bank of America, Nvidia is the best positioned company across the entire spectrum of Artificial Intelligence, not just because of GPUs.
Beyond everything already mentioned, its strategy includes investments and acquisitions in networking, inference, enterprise AI, workload management software, and full alignment with the supply chain for wafers and memory.
Analysts believe that with the next generation of large language models trained on Rubin and Blackwell, Nvidia will further increase its market share against competitors such as Google’s TPUs. They consider Nvidia the most complete bet in AI.
So for anyone who believes in progress, technology and the future of Artificial Intelligence, the conclusion is simple. Invest in Nvidia.