
At COMPUTEX 2026, NVIDIA officially introduced its latest major consumer computer chip, the RTX Spark.
The RTX Spark is not just a standard processor. It is a superchip that combines two parts. First, an NVIDIA Grace CPU with up to 20 cores. Second, an NVIDIA Blackwell GPU with 6,144 CUDA cores. That GPU is essentially an RTX 5070, the non Ti version. The chip comes with up to 128GB of fast LPDDR5X unified memory, which means the CPU and GPU share the same memory pool. It is built on TSMC’s 3nm manufacturing process.
The technology is based on NVIDIA’s GB10 chip, which is used inside the company’s DGX Spark platform. That platform is a line of small, powerful desktop computers aimed at businesses and runs a custom version of Ubuntu Linux.

Speaking at the event, NVIDIA CEO Jensen Huang said the main goal is not just gaming or video editing, though the chip works well for both. Instead, he focused on running AI agents directly on a local computer. AI agents are the next step beyond large language models.
To do this, a machine needs both strong processing power and lots of memory. That is where the 128GB of unified memory comes in. NVIDIA claims the RTX Spark can deliver one petaflop of performance for AI workloads.
The RTX Spark is designed for thin, powerful laptops. These devices are expected to arrive in the fall of 2026. NVIDIA has partnered with several major computer makers, including Acer, Asus, Dell, Gigabyte, HP, Lenovo, MSI, and Microsoft.
This article, NVIDIA’s RTX Spark brings 20 cores, 128GB RAM, and 1 petaflop of AI performance to laptops, was originally published at NoypiGeeks | Philippines Technology News, Reviews and How to's.
Comments
Post a Comment