Skip to the main content block
::: Home| Sitemap| Podcasts|
|
Language
Featured Programs
繁體中文 简体中文 English Français Deutsch Indonesian 日本語 한국어 Русский Español ภาษาไทย Tiếng Việt Tagalog Bahasa Melayu Українська Sitemap

Nvidia, MediaTek unveil co-developed RTX Spark made for AI agent computing

02/06/2026 15:42
Editor: Tristan Hilderbrand
Marvell CEO Matt Murphy (right) gave a keynote speech this Tuesday, June 2, 2026, sharing the stage with Nvidia CEO Jensen Huang (left). (Photo: CNA)
Marvell CEO Matt Murphy (right) gave a keynote speech this Tuesday, June 2, 2026, sharing the stage with Nvidia CEO Jensen Huang (left). (Photo: CNA)

Nvidia CEO Jensen Huang on Tuesday praised MediaTek's engineering capabilities as he appeared alongside the company’s CEO Rick Tsai (蔡力行) to discuss the jointly developed RTX Spark processor for Windows computers.

Speaking at a global media briefing, Huang said the project brought together hundreds of engineers from both companies and took nearly two years to complete.

The personal computer industry is undergoing its biggest transformation in four decades, Huang said, claiming that computers are evolving from traditional computing tools into platforms capable of running artificial intelligence agents with autonomous execution and reasoning abilities.

Huang said discussions with Microsoft about future generations of the Windows platform prompted Nvidia to consider what type of chips and computer architecture would be needed for AI agents. They decided to create RTX Spark, a system-on-chip platform designed to deliver high performance with low power consumption.

The platform incorporates Nvidia’s NVLink interconnect technology and is compatible with both Windows and the CUDA ecosystem. The Nvidia CEO said it is designed to support AI agent workloads, industrial applications, and gaming.

Tsai said the collaboration was challenging but successful. He noted that MediaTek's expertise in low-power CPU technology, combined with NVLink's high-speed connectivity and a unified memory architecture linking CPUs and GPUs, helped create what he described as an “ideal platform for AI agent computing.”

為提供您更好的網站服務,本網站使用cookies。

若您繼續瀏覽網頁即表示您同意我們的cookies政策,進一步了解隱私權政策。 

我了解