NVIDIA RTX Spark Laptops Officially Launch, Finally Offer Windows Users a Strong MacBook Pro Alternative

NVIDIA RTX Spark machines reached store shelves yesterday from ASUS, Dell, HP, Lenovo, Microsoft, and MSI, with the Surface Laptop Ultra and ProArt models among the first available starting at $2,599. These systems pack a single superchip that combines a custom Grace CPU designed with MediaTek and a Blackwell RTX GPU into one package, sharing up to 128GB of LPDDR5X memory so both the processor and graphics draw from the same pool without the usual split between system RAM and dedicated video memory.

That memory configuration is at the heart of what these laptops can do. Top-end models use a 20-core Grace design, 10 high-performance Cortex-X925 cores, and 10 efficiency A725 cores, together with a Blackwell GPU with 6,144 CUDA cores. In terms of shader count, that is about equivalent to an RTX 5070. Lower-end versions are limited to 18 cores and 5,120 shaders. Memory bandwidth is roughly 300GB/s, and NVIDIA claims that this technology can give up to one petaflop of FP4 AI performance while using only 45-to-80 watts of power. Tests have already proven that the GPU can compete with Apple’s M5 Max in some rendering and AI applications, despite the fact that the single-core and multi-core CPU numbers remain far lower than the M5 Pro and Max.

The available partner laptops range in size from 14 to 16 inches and include either OLED or mini-LED displays, with several being touch enabled and capable of 120Hz or higher refresh rates. The ASUS ProArt P16 features a 4K OLED screen with all the bells and whistles, as well as complete USB-A connectivity, whilst the Microsoft Surface Laptop Ultra has a more modest 15-inch mini-LED display and a good set of USB4 connections. Most models weigh between 1.5 and 2.2 kg, with a few also featuring stylus support or 2-in-1 hinges. Storage capacity is limited to 2TB on many configurations, and early hands-on reviews have praised the metal construction, haptic trackpads, and speakers, even if they can get noisy and warm under extended loads.

Local AI work is most likely where the unified memory and CUDA stack shine the brightest. If you have a laptop with 64GB or 128GB of memory, you can run large language models that would otherwise necessitate extensive memory management or cloud offloading on other devices. Adobe apps benefit from TensorRT acceleration for faster effects and rendering in beta builds, and some tests have even shown that a Surface Laptop Ultra outperforms an M5 Max MacBook Pro on some inference runs, despite the fact that other creative suites still prefer Apple’s more mature software and higher memory bandwidth. Gaming benefits from DLSS 5, ray tracing, and Reflex support, and day one titles such as Cyberpunk 2077 and Fortnite should perform similarly to those on an RTX 5070 laptop.

The battery life is perhaps the largest drawback when compared to a MacBook Pro machine. Web browsing testing on the Surface Laptop Ultra lasted about 10 hours, which is nearly half the time of an M5 Pro 16-inch machine under similar conditions. And it only gets worse when you start operating large AI or gaming sessions. Performance suffers significantly when the laptop is powered by batteries, more so than NVIDIA claimed. Windows on Arm still requires some emulation for older software, even though most of the major creative and productivity apps now operate natively, or at least without much difficulty. To top it all off, anti-cheat technology is still lacking in some multiplayer games, limiting your gaming options when compared to a standard Windows laptop.

Pricing noticeably brings the base price points in line with mid-tier MacBook Pro configurations, while the higher spec models occasionally undercut the fully loaded M5 Max machines. A 128GB Surface or ProArt build costs less than an equivalent MacBook Pro with the same level of memory and storage, but it sacrifices Apple’s typically longer battery life, superior single core performance, and more established ProRes and Final Cut workflows. Instead, users gain CUDA libraries, improved game compatibility, and the ability to run monster AI models on their local machines rather than relying on the cloud. For developers and creatives who are already immersed in NVIDIA’s software offerings or want access to Windows-specific capabilities, these laptops fill a void left by prior high-end Windows laptops.