Search  for anything...
ASUS

ASUS Ascent GX10 AI Supercomputer, DGX Spark, NVIDIA GB10 Superchip, 128GB LPDDR5x, 1TB PCIe Gen4 NVMe SSD, Wi-Fi 7 & BT5.4, Agentic AI Ready, Supports OpenClaw, NemoClaw, Stackable Chassis

  • Based on 62 reviews
Condition: New
Checking for the best price...
$3,970.99 Why this price?
Save $4,206.01 was $8,177.00

Buy Now, Pay Later

As low as / mo
  • – Up to 36-month term if approved
  • – No impact on credit to apply
  • – Instant approval decision
  • – Secure and straightforward checkout

Ready to go? Add this product to your cart and select a plan during checkout.

Payment plans are offered through our trusted finance partners Klarna, Affirm, Afterpay, Zip, Apple Pay, and Google Pay. No-credit-needed leasing options through Acima may also be available at checkout.

Learn more about financing & leasing here.

Selected Option

Free shipping on this product

FREE 30-day refund/replacement

To qualify for a full refund, items must be returned in their original, unused condition. If an item is returned in a used, damaged, or materially different state, you may be granted a partial refund.

To initiate a return, please visit our Returns Center.

View our full returns policy here.

Availability: In Stock.
Fulfilled by Amazon
Arrives Monday, Jul 27
Order within 17 hours and 19 minutes
Available payment plans shown during checkout
Protection Plan Protect Your Purchase
Checking for protection plans...

Recently Viewed

Size: 1TB PCIe 4.0 NVMe SSD


Style: NVMe SSD


Features

  • Extreme AI Performance: Powered by NVIDIA GB10 Grace Blackwell Superchip delivering 1 petaFLOP of AI performance and 128GB memory for 200B model fine-tuning.
  • Developer-Optimized Platform: Designed for AI developers building secure, long-running agentic workflows, with compatibility across frameworks such as OpenClaw and NemoClaw, supporting private on-device inference, sandboxed execution, and governed data access.
  • Scalable Architecture: Featuring NVIDIA NVLink-C2C for ultra-fast CPU-GPU memory communication and NVIDIA ConnectX-7 networking to support dual GX10 system stacking, unlocking superior scalability and performance.
  • Advanced Thermal Design: Engineered cooling ensures sustained high performance and reliability in an ultra-small form factor.
  • Full Stack AI Solution: The GB10 and NVIDIA AI software stack provide a full stack solution for AI development and deployment.

Description

Behold the ASUS Ascent GX10 Personal AI Supercomputer, a revolutionary force echoing the power of NVIDIA DGX Spark with its cutting-edge NVIDIA GB10 Superchip. Unleashing petaflop-scale AI on your very desktop, this marvel ignites the potential of developers, AI researchers, and data scientists, fostering local AI breakthroughs with unparalleled performance and ingenuity.

Brand: ASUS


Operating System: NVIDIA DGX OS


CPU Model: Cortex


CPU Speed: 3.8 GHz


Cache Size: 128 GB


Graphics Card Description: Dedicated Nvidia Graphics


Graphics Coprocessor: Blackwell GPU


Memory Storage Capacity: 128 GB


Specific Uses For Product: AI Development, AI Research, Machine Learning


Personal computer design type: Mini PC


Operating System: NVIDIA DGX OS


Specific Uses For Product: AI Development, AI Research, Machine Learning


Personal Computer Design Type: Mini PC


Color: Stellar Gray


Additional Features: 128GB RAM, NVIDIA ConnectX-7 SmartNIC, NVIDIA GB10 Superchip, Stackable Magnetic Feet, Ubuntu Linux OS


Hard Disk Description: SSD


Hardware Interface: Bluetooth 5, Ethernet, HDMI, USB 3.2 Gen 2×2


Power Consumption: 240 watts


Item Dimensions: 5.91 x 5.91 x 2.01 inches


Item Weight: 3.3 pounds


Video Output Interface: HDMI


Hard Disk Interface: PCIE x 4


Style Name: NVMe SSD


Cooling Method: Fan


Compatible Devices: Keyboard, Monitor, Mouse


Power Plug Type: Type B - 3 pin (North American)


Graphics Description: Dedicated Nvidia Graphics


Graphics Coprocessor: Blackwell GPU


Graphics Card Ram: 128 GB


Graphics Ram Type: Shared


Graphics Card Interface: PCI Express


Processor Series: Cortex


Processor Speed: 3.8 GHz


Processor Socket: Information Not Available


Processor Count: 1


CPU Model Speed Maximum: 4 GHz


Total Usb Ports: 4


Total Number of HDMI Ports: 1


Human-Interface Input: Buttons


Keyboard Description: No physical keyword


Keyboard Layout: QWERTY


Brand: ASUS


Model Number: GX10-GG0010BN


Model Name: GX10-GG0010BN


Built-In Media: GX10-GG0010BN, Power Adapter, Power Cord


Processor Brand: NVIDIA


Model Year: 2026


CPU Model Number: GB10


Warranty Description: 1 Year Manufacturer


Video Processor: NVIDIA


UPC: 199291183124


Manufacturer: ASUS


Cache Memory Installed Size: 128 GB


Memory Storage Capacity: 128 GB


RAM Memory Installed: 128 GB


RAM Memory Technology: LPDDR5x


Ram Memory Maximum Size: 128 GB


Memory Speed: 6400 mt_s


RAM Type: DDR5 RAM


Memory Clock Speed: 8533 MHz


Display Resolution Maximum: 3840x2160 pixels


Display Type: No Display


Aspect Ratio: 169


Resolution: 4K


Native Resolution: 3840 x 2160


Connectivity Technology: 10G LAN, Bluetooth 5.4, HDMI, USB, Wi-Fi 7


Wireless Compability: 802.11.be, Bluetooth


Wireless Technology: Bluetooth, Wi-Fi


Frequently asked questions

If you place your order now, the estimated arrival date for this product is: Monday, Jul 27

Yes, absolutely! You may return this product for a full refund within 30 days of receiving it.

To initiate a return, please visit our Returns Center.

View our full returns policy here.

  • Klarna Financing
  • Affirm Pay in 4
  • Affirm Financing
  • Afterpay Financing
  • Zip Pay in 4
  • Financing through Apple Pay
  • Financing through Google Pay
Leasing options through Acima may also be available during checkout.

Learn more about financing & leasing here.

Similar Products

Top Amazon Reviews

    • Hidden Gem and The Best of The Lineup
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
I want to say that the review by Regular Person hit the nail on the head, and I totally forgot about running this headless, and now I am. Thank you. I want to add to their review by saying that the negative reviews by influencers early on are a blessing in disguise if you are trying to look at other computers for anything shared memory 128gb+. This is available and I feel lucky to have picked not only a GB10 but my favorite brand ASUS. No one is talking about the build quality of ASUS. All I can say is do your own research and see that ASUS built this from scratch, it is not reference with a few extra things. I am a fan of the ASUS TUF line and ROG. This is because they use top notch parts/pieces that are military spec and last longer than you need. This unit is exactly that. They designed and made their own board, and the parts are all high quality for this unit to be certified MIL-STD 810H. The cooling is also all their own design. Don't take my word, look it up. I'm not in IT or computer science, I'm an enthusiast. If you are like me, then I highly recommend having something like Google Gemini Pro ready when you begin the startup. My hand was held the entire time during the setup, and let me tell you, I could not have set this up the way I wanted without using another AI for all my dumb questions. I almost had a heart attack on the very first major update when it was supposed to reboot and didn't. Gemini told me to wait up to 25 minutes before pressing the power, and I followed the recommendations to the T. I am super excited to begin my new supercomputer journey with Hermes, Deepseek as of today and ready to start building and sharing my journey for others to see. I have never seen technology advance like AI is doing and developing faster and faster, so fast I moved beyond subscriptions, and my video card opensource models. Hopefully I can make this profitable, as that is what I told my wife in order to justify the purchase. Enjoy everyone! ... show more
Reviewed in the United States on May 14, 2026 by Eric
    • Reliable and stable, getting better every week
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
I think it is important to note that many reviews on the Grace Blackwell machines that are from late 2025 or early 2026 may not apply anymore. This runs the Nvidia Ubuntu operating system, and Nvidia has really been great about fixing a lot of problems the system had in running this very different architecture when it was first released. This is my 2nd system. These systems are getting more and more pleasant to own and use every week - support for them is growing in the opensource world, and NVIDIA is pushing out docker containers that are ready made to do many things well if you choose that path- they offer playbooks with easy step by step instructions that make it seem more like you are getting great software custom made for these new architectures in addition to a great machine for the price you pay. I was a bit disappointed that adding the $$$ cable and clustering 2 together didn't really give me the one combined machine I'd hoped for, but I do like my current set up. (It worked fine,and let me run a huge model, but it just wasn't optimal for my use to do so.) On this machine, I am only using it for inference. I am running VLLM on it, currently with Qwen 3.6 31B but it's always available. I have ran larger models on it, but this seems to be the sweetspot for me with the agentic systems that connect to it. I hardly ever go past 65% system memory I have room to grow but system availability is WONDERFUL if I don't consider reboots due to system upgrades. Do expect FREQUENT UPDATES. Nvidia has been pushing updates almost daily - sometimes multiple times a day, and they all require reboot. The NVIDIA sync program works great for this machine. I run mine headless - you can set it up headless too if you like. I did, and didn't have a bit of trouble. I do suggest that you go ahead and make a repair boot-able USB disk - you can get that on the NVIDIA site. This is NOT regular Ubuntu. I am disappointed by people posting negative reviews that buy these systems for gaming. They are NOT for gaming natively. These are AI machines. This will keep your inference in-house if that's what you want - one is enough to power a small business with inference with a model that is good at tool call, and vision. Whether or not you should get a 1tb or a 4tb one depends a bit on how you are going to use it. I have one with 4tb that I use for storing multiple models, but I got this one with 1tb because it is only ever going to run 1 model and vllm. 1tb is more than enough to store a few models and vllm. It depends on your use case. I'd get the 4 if I wanted to do more like generative work with diffusion models, comfy-UI/voice/ music. I do those things on my 4tb and run Ollama too, but if you are going to run multiple services like that, even though you have the HUGE combined memory pool, you still have to think about how you are going to allocate and time resource use. The graphics look great running generative text or image to video. Video is smooth as can be. It handles heat well - but inference is HOT. I highly recommend that you do NOT point the backside at you - it can be like having a small space heater right next to you. I learned that the hard way.I have mine on a stainless steel commercial kitchen table and that is VERY helpful if you have more than one of these running. It's not as bad if they are not always available. Just remember, inference is HOT. If they get too hot, they will go into thermal throttling. I find they really do need a cool room, and good airflow for optimal operation. There's more than enough ports, but they are a bit tight. I do like the power button on the front. They use a regular Ethernet port, and have wireless capability built in. ... show more
Reviewed in the United States on May 1, 2026 by Regular Person
    • Only for researchers - not inference, not small scale fine tuning
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
I thought this box would be the solution to the Mac studio drought, but it was far from it. Inference is bottlenecked by decoding which is extremely slow for this device's memory bandwidth. I gave up on inference and started using it fine tune. While it is low profile / noise / power, speeds of small scale fine tuning were outpaced by a RTX3090. Depsite the high compute, there is no good way to utilize them as a consumer focused on small scale fine tuning and LLM inference. I tried to explore all the options to utilize the GX10, but none of the performance for my use cases justified the price tag. It is a cool device and useful for researchers working on Blackwell development and prototyping, but it was not a good fit for me and probably a lot of people looking for consumer local/private AI for work. At least not worth this price. ... show more
Reviewed in the United States on May 19, 2026 by Jackson Lim
    • Good for local inference
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
It's only been a week, but I love it. I managed to automate some development tasks and it ran several days without needing my intervention. According to it's own reported energy consumption it is efficient, even if produces only a few tokens per second, but it is consistent with the reviews I saw online before buying it. I imagine people buying a computer like this will do some research beforehand, and I think you get exactly what you expect. The only thing I can't talk about is durability. ... show more
Reviewed in the United States on May 16, 2026 by Alexei Eleusis Díaz Vera
    • Only for researchers
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
First, be aware the seller charges a 10% restocking fee. $300 plus $160 for insurance from UPS to send it back was a lot of money to find out that the DGX Spark does not live up to the hype for my use case. Overall, the hardware is good. I didn’t hear a whisper of the fan in 3 days of testing either. I have read that the Nvidia version is loud. I would recommend the Asus Ascent for its design. My problem is with the software. DGX OS is easy to set up, but it is locked into CUDA 13. I first tried to run the newest Nemotron Nano using NIM, but it has a requirement for CUDA 13.1. After upgrading, I kept getting so many errors indicating a mismatch between CUDA and the OS that I reinstalled the OS, which took an hour. Next I tried GPT-OSS:120B running under Ollama, and while it worked, I got roughly 10-15 tokens per second. NIM did work, but it was not any faster. So this supercomputer which is supposed to be able to handle 200 billion parameter models is very slow if you can get it to work. Depending on your usage, if you just want to do inference, any cloud service will take years to equal the cost of the DGX Spark. My RTX 5090 runs Nano and GPT-OSS:20B at 100-200 tokens/s, and it was the same cost. It has the added benefit of running Cyberpunk 2077 at the highest settings at 100+ fps too. This review is solely on the models mentioned. YMMV for other models or use cases. If you buy a Spark, the Asus Ascent is a good choice. ... show more
Reviewed in the United States on January 10, 2026 by Morris L.
    • Best value for this platform right now.
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
Review Image by Scott C.
This is one of the best ways to adopt this platform since SSD prices are so high right now. The SSD can be cloned to a larger one and installed in the one slot later if you need to. The power brick is wide but only about an inch high. The power brick is warm anytime the unit is on.
Reviewed in the United States on May 13, 2026 by Scott C.
    • Nvidia Officially Doesn’t Support GB10
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
Nvidia does not support these. They will push driver updates without any announcements or alerts and if you restart during the update it will brick the GPU. Nvidia support - what you’re paying for with one of these - will close your ticket and tell you to ask for help from the community forums. My company tested these for a data center build out: if they won’t support these you should expect poor support on their full racks. We’re going with AMD instead - their dev stack isn’t as robust but we actually get support. The only real drawback has been that we can’t just drop one of these on every developers desk - it all runs on big iron, but that’s mostly an architecture problem that only causes problems if you need remote asynchronous non-networked developers. It’s only a few hundred million on our deal, but still an easy choice. When these actually worked they were pretty good though - so if you can afford to have $4k throw away devices, give it a try. ... show more
Reviewed in the United States on June 6, 2026 by Robert Beckman
    • Awesome cooling
Size: 1TB PCIe 4.0 NVMe SSD Style: NVMe SSD
Pretty good, high load at 50deg c, prefill is one of the highest any homelabber can buy, don’t see review about single use LLM token/s, that is not how agentic works. This is why I’m buying the 2nd one.
Reviewed in the United States on July 13, 2026 by alan g.

Can't find a product?

Find it on Amazon first, then paste the link below.
Checking for best price...