ASUS Unveils Ascent GX10 Personal AI Supercomputer – Petaflops of AI Power in the Palm of Your Hand
ASUS breaks new ground in AI computing with the launch of the ASUS Ascent GX10, a personal AI supercomputer that packs up to 1 petaflop of processing power into a compact footprint of just 150 x 150 x 51 mm.
Powered by the NVIDIA® GB10 Grace Blackwell Superchip and the integrated NVIDIA AI Software Stack, it will be available globally on October 15, 2025, and will hit the Thai market in November 2025.
Superior power from NVIDIA GB10 Grace Blackwell Superchip
At the heart of the ASUS Ascent GX10 is the NVIDIA GB10 Superchip , which combines a 20-core NVIDIA Grace CPU with an NVIDIA Blackwell GPU in a single chip.
Delivers up to 1 petaflop of AI inference and model fine-tuning performance with 128GB of Unified Memory.
Direct support for working with large models up to 200 billion parameters — making the GX10 the perfect choice for researchers, AI developers, and data scientists who need high-performance AI without the need for the cloud.
A complete AI software stack – ready to use
ASUS Ascent GX10 comes with the NVIDIA AI Software Stack , which includes all the tools for AI work, from
-
Prototyping
-
Fine-tuning (Model tuning)
-
Inference (prediction)
It covers developments in many fields such as Robotics , Computer Vision , and Vision-Language Models (VLMs).
It enables users to quickly and efficiently start developing and testing AI models.
Scalable Architecture – Double the Power with Dual GX10
Ascent GX10 is designed for scalable interconnectivity (Scalable Architecture).
Via the high-speed NVIDIA ConnectX-7 interface and NVLink-C2C technology for ultra-fast communication between the CPU and GPU.
When two Ascent GX10s are connected together, performance increases to 2 petaflops.
With up to 256GB Unified Memory and up to 8TB of storage.
Perfectly meets the needs of large-scale AI training within organizations
Intelligent cooling system – 1.6 times more efficient
Despite its compact size, the Ascent GX10 is meticulously engineered.
With Dual Fan cooling system size 140 x 80 mm.
5 Heat Pipes , and 7-level intelligent fan control
Helps increase cooling efficiency by 1.6 times compared to similar sized models.
With quiet and stable operation even under intensive AI workloads 24/7.
Scalable storage for AI tasks
The ASUS Ascent GX10 has SSD options to suit every level of usage.
-
1TB PCIe 4.0 SSD: Ideal for AI beginners or small projects.
-
2TB PCIe 4.0 SSD: For workloads that handle multiple models or large data sets (coming soon)
-
4TB PCIe 5.0 SSD: For maximum performance for advanced research or AI projects (coming soon)
All models support upgrades and secure internal data storage with Enterprise-level Security.
Small but powerful – AI in the palm of your hand
Weighing only 1.48 kg and compact size
The Ascent GX10 can be installed in virtually any workspace — from research desks to AI labs.
Equipped with complete connectivity such as USB-C® Gen 2x2, HDMI 2.1, LAN 10G, and Wi-Fi 7 (Gig+).
Summary Specifications
| list | details |
|---|---|
| Processor | NVIDIA GB10 Grace Blackwell Superchip (ARM v9.2-A, 20-core) |
| GPU | NVIDIA Blackwell (Integrated) |
| Memory | 128GB LPDDR5x Unified Memory |
| Storage | 1TB / 2TB / 4TB NVMe PCIe SSD |
| Network | 10G LAN, Wi-Fi 7, Bluetooth 5 |
| OS / Software | NVIDIA DGX OS, NVIDIA AI Software Stack |
| Size / Weight | 150 × 150 × 51 mm / 1.48 kg |
| Power | Up to 240W (USB-C PD 3.1 EPR) |
Available worldwide October 15, 2025 – See you in Thailand this November.
ASUS has confirmed that the Ascent GX10 Personal AI Supercomputer will be available globally starting October 15, 2025.
and there is a plan to officially launch in Thailand in November 2025
This marks another important step in making enterprise-level AI computing technology more accessible to individuals.
📍 More details:
👉 ASUS Ascent GX10 Official Website
👉 ASUS Business
👉 ASUS LinkedIn
summarize

ASUS Ascent GX10 marks the beginning of a new era of “personal AI”.
Combining supercomputer- level processing power with a compact size, it can be used anywhere.
Ideal for AI developers who need speed, flexibility, and security for local operations.
Ready to support the future of AI that grows without limits