Asus Ascent GX10 AI Nvidia Grace Blackwell GB10 128GB DDR4 1TB SSD Ultra-small Super Computer
The ASUS Ascent GX10 is ASUS's version of NVIDIA's DGX Spark reference design — a compact desktop AI supercomputer built around the same NVIDIA GB10 Grace Blackwell superchip that powers PNY's DGX Spark elsewhere in this range and NVIDIA's own reference unit, pairing a 20-core Arm CPU with a Blackwell-architecture GPU on one package for up to roughly 1 petaFLOP of FP4 AI compute. ASUS is one of a small number of NVIDIA's official partners building this exact reference platform, so the core compute hardware is effectively identical across all these OEM variants — what differs between vendors is industrial design, included accessories, and support/warranty terms, not the underlying chip or memory architecture.
It's configured with 128GB of unified system memory shared coherently between CPU and GPU and a 1TB SSD in this listing. Worth flagging honestly: the catalog name lists the memory as 'DDR4,' but NVIDIA's GB10 platform (and every DGX Spark-class device, including ASUS's own) uses LPDDR5x unified memory, not DDR4 — this reads as a labelling inconsistency in the product name rather than a genuine spec difference from the platform this device is built on. Like every DGX Spark-class device, it runs NVIDIA's DGX OS with the full AI software stack preinstalled, and is built for local AI model development, fine-tuning, and inference — running and experimenting with large models on a desk rather than renting cloud GPU time, in a genuinely tiny desktop-sized enclosure.
Online price incl. VAT
Availability: Stock Pending
Compute
- Chip: NVIDIA GB10 Grace Blackwell Superchip
- CPU: 20-core Arm (10x Cortex-X925 + 10x Cortex-A725)
- AI Performance: Up to ~1 PFLOP FP4
Memory & Storage
- Memory: 128GB unified memory (listed as DDR4 in the product name; the GB10 platform actually uses LPDDR5x — flagged, not resolved)
- Storage: 1TB SSD
General
- OS: NVIDIA DGX OS
- Form Factor: Ultra-compact desktop AI computer
- Use Case: Local AI model development, fine-tuning, inference