ASUS Ascent GX10 introduced – Features, performance & possible uses in the AI ​​sector

Supercomputers in the age of AI: Why we need more computing power

Supercomputers have traditionally represented the pinnacle of computing power—machines that fill entire halls and were previously used only in national research centers or for highly complex simulations. But with the breakthrough of artificial intelligence (AI) and large language models (LLMs), the picture has changed. Today, not only universities but also companies require access to massive computing power.

Training and deploying models with hundreds of billions of parameters poses challenges even for large cloud infrastructures. At the same time, the need to run AI systems on-premises is growing , whether for data protection reasons or to better control costs and latency.

With the ASUS Ascent GX10 , also known as NVIDIA DGX Spark , ASUS is launching a new system that addresses precisely this need. Equipped with the latest NVIDIA Grace Blackwell architecture, the GX10 combines extreme performance with high energy efficiency – offering researchers and businesses a platform for the next generation of AI applications.

Performance, storage & network – the power of the GX10 in detail

At the heart of the GX10 is the NVIDIA GB200 Grace Blackwell superchip . This GPU and CPU combination is specifically designed for AI workloads and offers a tight integration of computing power and memory bandwidth. The GPU units deliver up to 1 petaflop of AI computing power in FP16 format —enough to train or run even the largest LLMs like GPT-4 or models with over 400 billion parameters.

The ARM-based Grace CPU complements the GPU with high energy efficiency and bandwidth. The combination creates a system that enables not only pure performance but also optimized data processing. This is supported by HBM3e memory , which allows for rapid access to training data of several terabytes.

For storage and data management, ASUS relies on NVMe SSDs , which can handle even large data sets without bottlenecks. With modern networking standards such as 400G Ethernet and NVIDIA Quantum-2 InfiniBand, the GX10 can also be easily integrated into clusters or linked with other systems.

The system is also well-positioned on the software side: It comes with DGX OS and supports common frameworks such as PyTorch, TensorFlow, and Hugging Face. This allows developers to get started right away without having to make complex system adjustments.

criterion	ASUS Ascent GX10 (NVIDIA DGX Spark)	NVIDIA DGX H100	Cloud (AWS/Azure/Google)
Computing power (AI FP16)	up to 1 PFLOP	approx. 0.5–0.7 PFLOP	Variable (depending on instances)
Maximum model size	200B – 400B+ Parameter	up to ~175B parameters	Any, depending on costs
Energy efficiency	Very High (Grace Blackwell)	Good, but less than GB200	Medium to high (depending on hardware)
Data sovereignty	Completely on-premises	Completely on-premises	Data stored externally
Investment cost	High (one-time)	High (one-time)	None (Pay-as-you-go)
Running costs	Low (power, cooling)	Low (power, cooling)	High (ongoing, often expensive)
Scalability	Very good, multiple systems can be linked	Good, cluster possible	Very high, virtually unlimited

Rack-compatible, robust and efficient: GX10 design

The ASUS Ascent GX10 is designed as a rack-mountable device and integrates seamlessly into existing data center infrastructures. The ASUS Ascent GX10's cooling system is based on a powerful liquid cooling system specifically designed for continuous operation under full load. This keeps the system stable, quiet, and energy-efficient even under high-intensity workloads.

A special feature is its expandability . Two GX10 systems can be coupled, opening up the possibility of training even larger models or performing inference tasks on a very large scale. This is a crucial option for companies that want to grow long-term.

In addition, the system offers modern interfaces such as PCIe Gen5, multiple high-speed network connections, and convenient management ports. In addition to the device itself, the package also includes the pre-installed operating system, remote management tools, and a comprehensive support package, allowing users to get started quickly.

From research to business: Where the GX10 shows its strengths

The GX10 has a wide range of applications.

In research, it enables the training of highly complex models – for example, for language processing, genomics, or simulations in physics. Universities and research centers benefit not only from the performance but also from the ability to process data in-house.

The GX10 also offers companies a powerful alternative to the cloud. Whether building agentic AI applications , automating internal processes, or developing customized LLMs – the platform is flexible in its use.

Its use in data-sensitive scenarios is particularly exciting . Banks, hospitals, and public institutions often need to ensure that sensitive data never leaves their premises. With an on-premises supercomputer like the GX10, they retain full control over their data without having to forgo AI innovations.

Last but not least, the scalability is impressive . Those who start small can later expand their system by combining multiple GX10 units – thus building their own modular AI data center.

Who benefits from the GX10 – and what are its limitations?

With the ASUS Ascent GX10, ASUS has launched a supercomputer clearly tailored to the demands of modern AI. Thanks to the NVIDIA GB200 Grace Blackwell superchip, it delivers not only raw computing power but also an architecture optimized for handling massive amounts of data.

The GX10 is an exciting solution for research institutions and companies with high data and computing needs . It particularly demonstrates its strengths in scenarios where data cannot be outsourced or where maximum performance is required.

The GX10 demonstrates: "The future of supercomputers is more compact, more efficient – ​​and thus within reach of more and more organizations."

ASUS Ascent GX10 Supercomputer – Prices, Availability & Advice

Want to find out if the ASUS Ascent GX10 is the right choice for your business or research?
Visit our product page and request details .
Alternatively, we also offer individual consulting and support in setting up customized AI infrastructures.