Gigabyte unveils Nvidia Spark-powered AI TOP AERO desktop
The new Windows-based PC aims to bring high-end local AI development and content creation to personal workstations using Nvidia's Blackwell architecture.
The short version
- The AERO features a 20-core Arm-based Grace CPU and a Blackwell-architecture RTX GPU.
- It supports up to 128GB of unified LPDDR5X memory and a 4TB PCIe Gen5 SSD.
- The device is designed for local AI development, gaming, and content creation.
- Gigabyte has not yet released official pricing, configurations, or availability dates.
Why it matters
This hardware represents a shift toward 'personal AI computers' capable of running complex models locally rather than in the cloud. For creators, it means more privacy and faster iteration on AI tools like image generators and local LLMs.
A new standard for personal AI
Gigabyte recently unveiled the AI TOP AERO, a compact desktop designed specifically for local AI workloads. The machine is built on Nvidia's Spark platform and utilizes the Blackwell architecture for its RTX GPU. This hardware is intended to serve as a 'Personal AI Studio' for users who need high performance without relying on cloud services.
The device is marketed toward content creators, gamers, and developers who want to run models locally. By focusing on local execution, the AERO allows users to work with sensitive data more securely. It also provides a smoother experience for real-time applications like image generation and local language model interaction.
While the hardware is officially announced, Gigabyte has not yet provided specific pricing or a way to purchase the unit. The company currently directs interested buyers to check local announcements for regional details. This leaves the AERO as a high-spec concept that is currently awaiting its commercial rollout.
Technical specs and architecture
Under the hood, the AERO combines several high-end components into a roughly one-liter chassis. It features a 20-core Arm-based Grace CPU paired with a Blackwell-architecture RTX GPU. The system also includes up to 128GB of unified LPDDR5X memory, which allows the CPU and GPU to share the same pool of high-speed data.
Storage for the device is handled by an SSD that can reach up to 4TB using the PCIe Gen5 interface. The machine also includes standard connectivity options like USB Type-C, 10GbE, and Wi-Fi 7. Unlike some other Spark-based systems, the AERO does not mention built-in networking hardware for clustering multiple units together.
The hardware is closely related to the Nvidia N1X Superchip, which features 6,144 CUDA cores and a 140W TDP. While the AERO is built for Windows 11, it shares silicon with the GB10 chip used in other Spark models. This shared architecture suggests that the AERO is part of a broader ecosystem of Nvidia-powered AI hardware.
Software and local AI tools
Gigabyte is positioning the AERO to work with several popular open-source and proprietary AI tools. The company specifically mentions support for LM Studio, Ollama, and ComfyUI. These are common platforms used by enthusiasts to run large language models and stable diffusion models on their own hardware.
The desktop also supports Nvidia’s Nemotron models and the Nvidia PAIR tool. PAIR is an open-source router that can distribute independent inference requests across different PCs on a local network. This suggests that while the AERO is a standalone unit, it can still participate in larger local setups.
The choice of Windows 11 as the primary operating system is a key part of Gigabyte's strategy. By using a familiar OS, the company aims to make high-end AI hardware accessible to a wider audience. This contrasts with other Spark-based machines that run Linux-based systems like the DGX OS.
Market context and pricing
The AERO enters a market where high-capacity memory is becoming increasingly expensive. Nvidia recently raised the price of its own 128GB DGX Spark to $6,950, a significant jump from its original launch price. This trend suggests that the cost of high-end AI hardware will remain high for the foreseeable future.
Other competitors in the space are already selling similar hardware at premium prices. For example, MSI offers an EdgeMesa N AI+ with 128GB of memory for $7,999. Meanwhile, Microsoft's 128GB Surface RTX Spark Dev Box is listed at $5,999. These figures provide a rough estimate of what the AERO might cost once it is released.
Industry experts predict that the current memory crisis could last at least another three years. This means that the AERO will likely be a luxury item for professional developers and serious creators. Buyers will have to weigh the benefits of local 128GB memory against the high upfront cost of the hardware.
Cooling and future questions
One of the primary technical questions surrounding the AERO is how it will manage heat. The device is designed to fit into a one-liter chassis while powering a GPU with a 140W TDP. Efficient cooling in such a small form factor is a significant engineering challenge that remains to be seen in real-world testing.
There is also uncertainty regarding how much of the 128GB of memory a standard Windows workload can actually utilize. While the memory is unified, different applications may have different limitations on how they can access that shared pool. Users will need to see benchmarks to determine if the hardware performs as expected for specific tasks.
Finally, it remains unclear if the top-tier configuration will be available before the end of the year. Until Gigabyte provides regional pricing and stock information, the AERO remains a high-spec promise. The next step for the company is to move from a specification sheet to a retail reality for consumers.
What to watch
- Official pricing and regional availability from Gigabyte
- Real-world thermal benchmarks for the one-liter chassis
- Comparison of Windows vs. Linux performance on the N1X Superchip
Try it
Check the LM Studio or Ollama websites to see what kind of local models you can run on your current hardware.
Sources and links
- ReportingMezha.net: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by Mezha.net · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
