Is a $7,000 laptop worth it if you want to run AI without the cloud?

The conflict surrounding ASUS's new ProArt P16 is simple: it offers elite performance for local AI workloads at a price that excludes the average consumer. ASUS has positioned the device as the first laptop to use NVIDIA's RTX Spark chip, a system-on-a-chip designed to run large language models directly on the device. The Engadget review published on October 23, 2026, confirms the hardware is capable but notes the platform is built for a niche of power-hungry creators and researchers rather than the general public.

The cost structure defines the product's market. The base configuration starts at $2,800, but the price climbs sharply as memory increases. The specific unit tested for the review contains 128GB of RAM and carries a price tag of $7,000. For comparison, a dedicated gaming laptop with similar graphics performance can be found for significantly less. The review points to an ASUS Zephyrus G16 equipped with an RTX 5070 and a 240Hz OLED screen, available at Best Buy for $2,124.

The hardware architecture marks a departure from traditional x86 processors from Intel or AMD. RTX Spark combines a 20-core Arm CPU with a Blackwell GPU and up to 128GB of unified memory. This design mirrors Apple's M-series chips. While the reviewer acknowledged that this creates a new branch of PC design, it will sit alongside existing x86 and Qualcomm Snapdragon X hardware rather than replacing them. The memory-heavy nature of the platform is unlikely to lower prices in a market where AI demand has already inflated RAM costs.

Performance benchmarks show a mixed result. In PCMark 10 Applications, the P16 scored approximately 3,000 points lower than an ASUS A16 equipped with a Snapdragon X2 Elite chip. It also trailed slightly in Geekbench 6 single-threaded CPU testing. However, graphics and AI tasks favored the new device. The P16 beat a MacBook Pro with an M5 Max by 4,700 points in Geekbench 6 GPU tests. In Procyon's AI Computer Vision 2 test, it scored 3,895 compared to Apple's 3,267. Its AI Image Generation score of 1,519 was 600 points higher than the Apple machine.

In practical use, the laptop handled demanding tasks smoothly. It ran the Qwen3.8 27B model, which requires 42GB of RAM, with memory to spare. Gaming performance was solid, achieving 70 fps in Control: Resonant at 1440p with ultra settings, medium ray tracing, and 4x frame generation. Lowering the settings increased the frame rate to 120 fps. However, compatibility remains a hurdle for gamers, as most titles rely on Microsoft's Prism emulator, which can reduce performance or cause issues. Native Arm64 builds are currently available for titles such as Fortnite, DOTA 2, and Marvel Rivals.

Physical attributes include a sturdy aluminum chassis and a 16-inch OLED display with 1,600 nits of brightness. The device weighs 3.9 pounds, making it lighter than the 4.7-pound Apple MacBook Pro. Ports include three USB-C connections, USB Type A, HDMI, a headphone jack, and an SD card reader, though an Ethernet port is missing. Battery life in the PCMark 10 Applications test reached ten hours and 51 minutes, falling short of the 15-plus hours seen in other premium laptops. The reviewer noted that performance did not drop when the laptop was unplugged.

The review concludes that AI professionals are the primary target audience who can justify the expense. For general users or gamers, the recommendation is to wait at least a year before considering the RTX Spark platform. The black finish is prone to fingerprints, and the fans generate noticeable noise during high-load tasks. NVIDIA states that the graphics hardware is roughly on par with the RTX 5070 mobile GPU while drawing less power.