Jensen Huang Drops the Bombshell: NVIDIA RTX Spark Launches, Arm Laptops Take on Intel Head-On

📡 AI News 2026-07-11 1 min read

NVIDIA unveils RTX Spark at Computex: 20-core Arm CPU, 6,144 CUDA cores, 128GB unified memory at 45-80W. Microsoft's Surface Laptop Ultra debuts as first partner. PC industry shaken.

💡 What You Will Learn

NVIDIA unveils RTX Spark at Computex: 20-core Arm CPU, 6,144 CUDA cores, 128GB unified memory at 45-80W. Microsoft's Surface Laptop Ultra debuts as first partner. PC industry shaken.

Today at Computex Taipei, Jensen Huang's keynote was packed with bombshells. Fresh off announcing the DGX Station for Windows (a trillion-parameter desktop supercomputer), he dropped a second curveball — RTX Spark. This marks NVIDIA's return to consumer CPUs after more than a decade, and they're coming in hot with an Arm architecture, unified memory, and desktop-class GPU — taking on Intel, AMD, and Qualcomm head-on. Microsoft has already thrown its weight behind the platform: the brand-new Surface Laptop Ultra is built entirely around RTX Spark, from chip to system. Let's unpack what makes this thing so formidable. RTX Spark (formerly codenamed N1X) is a complete system-on-chip with two dies connected via NVLink-C2C high-speed interconnect. On the CPU side: 20 custom Grace Arm v9.2 cores co-developed with MediaTek. On the GPU side: a Blackwell RTX GPU packing 6,144 CUDA cores — matching the desktop RTX 5070. It features up to 128GB of unified LPDDR5X memory shared between CPU and GPU, delivers 1 petaFLOPS (FP4) of AI compute, and sips just 45-80W in laptop form. In plain English: NVIDIA has crammed desktop-class discrete GPU performance into a laptop chip drawing under 80W.

Related Articles
2026-08-31
没完了吗?主板也要涨价了!我算了算,现在我装一台电脑比去年贵了至少三千往上
2026-07-13
AMD CPUs Surge 46%! Steam Survey: Intel Drops from 81% to 54% in Five Years
2026-08-14
AI Scheduling Assistant for Healthcare 2026: How Clinics Cut No-Shows

Written by our editorial team; tools listed here are tested or verified against public sources. Links point to official sites or GitHub repos for reference only — no paid placements.

💬 Comments (0)

No comments yet. Be the first!

Login to comment