Gotrade News – Nvidia announced its RTX Spark superchip series with the new N1X processor for Windows laptops at COMPUTEX Taiwan. CEO Jensen Huang also introduced Cosmos 3, an open AI world model for robots and autonomous systems.
The launch of these two products marks Nvidia’s entry into the Windows PC processor market and the physical AI infrastructure layer. Investors see the move as a direct challenge to Intel, AMD and Apple silicon in mobile computing.
Important points
- Nvidia’s N1X chips power Surface laptops and devices from HP, Dell, Lenovo, Asus, Acer, and MSI.
- Cosmos 3 was trained on 20 trillion tokens of multimodal data containing nearly 1 billion images.
- Nvidia has confirmed that the Vera Rubin AI platform is in full production, with shipping expected in the fall.
According to Investing.com, the N1X processor was developed in collaboration with Microsoft and designed by MediaTek on the Arm platform. Huang said these chips are primarily intended to run locally hosted artificial intelligence agents on consumer hardware.
The NVIDIA (NVDA) partner’s new laptops could rival Apple’s MacBook series, marking the company’s entry into a segment long dominated by Intel. Microsoft’s (MSFT) Surface laptops will be among the first devices to feature the new processor.
A direct challenge to Intel and AMD
The push for Windows puts Nvidia in direct competition with Intel (INTC) and AMD in the PC processor market. Analysts see the partnership with Microsoft as a structural change in the laptop silicon supply chain.
As reported by Investing.com, Huang also introduced another chip called Vera, which is specifically designed to run AI agents. The executive claimed that Vera can complete agent tasks 80% faster compared to existing technology.
Nvidia has confirmed that its next-generation Vera Rubin AI platform is now in full production. Product shipments for the platform are expected to begin in the fall of 2026, according to Investing.com.
Cosmos 3 targets physical AI
According to Axios, Cosmos 3 is an open-world model designed to help robots, autonomous vehicles, and other physical systems predict their real-world environments. Nvidia trained the model using 400 million real and synthetic videos, as well as ambient audio, text, and action data.
According to Ming-Yu Liu, vice president of Nvidia’s Cosmos Lab, Cosmos 3 is differentiated from traditional video generators by its action data layer. Liu told Axios that autonomous actions are key to modeling not only how a scene will look, but also how the machine will move.
Developers can use Cosmos 3 to simulate actions in a physical environment and build task-specific robot models on top of it. For each Axios, the model generates the robot’s joint angles, gripper positions, and trajectories to train the machine for real-world tasks.
Nvidia releases super model for high-precision training of robots and self-driving cars. As Axios reports, nano models can produce results in a fraction of a second, and edge models are coming soon.
The company’s broader bet is that the next wave of AI will need to predict, simulate, and act in the physical world. Nvidia wants its open model and infrastructure to become the default starting point for developers.
