SI Agent News
DevelopingChips

Gigabyte puts a small Nvidia-powered desktop for local AI on sale

Gigabyte's AI TOP AERO runs AI models on your own desk, not in the cloud. Here is what the company claims, and what it leaves out.

128GBmaximum unified system memory offered

The short version

Why it matters

Running AI on your own machine keeps data on the device and avoids paying per use in the cloud. This release shows hardware makers now selling that idea as a mainstream desktop. Everything here comes from the seller, so independent testing still has to confirm it.

What Gigabyte announced

Gigabyte has announced that its AI TOP AERO desktop is officially on sale, as PR Newswire reported. The company says it went on sale on November 10, 2026. It is built on Nvidia's RTX Spark platform and runs Windows 11.

Gigabyte calls it a "personal AI studio". The idea is one small box on a desk that can handle building AI software, making content and playing games, without sending your work to a remote data center.

The release does not give a price. It says configurations, prices and supply vary by region and tells buyers to check Gigabyte's local announcements.

What is inside the box

The machine uses an Nvidia chip called the RTX Spark N1X. It combines a 20-core Nvidia Grace CPU (the general-purpose processor) with a Blackwell RTX GPU (the graphics processor that does most of the AI math).

The memory is the headline spec. Gigabyte lists up to 128GB of LPDDR5x unified memory, meaning one shared pool the whole system draws on. For AI, memory size largely decides how big a model you can load, which is why the number matters. Storage goes up to 4TB on a PCIe Gen5 SSD, a fast type of drive.

For connections, the release lists USB Type-C, 10GbE wired networking and Wi-Fi 7. The case is about 1 liter, so cooling is the obvious challenge. Gigabyte says it built its own cooling design with extra thermal headroom, so the machine can hold its performance during long, heavy jobs. That claim is the kind that independent reviews exist to test.

What you could run on it

Gigabyte says developers can use Nvidia's CUDA platform, its software toolkit for programming its chips, to build, prototype, fine-tune and run recent open models locally. It also names Nvidia Nemotron models and says the machine works with tools such as LM Studio and Ollama, which are popular programs for running models on your own computer.

The release stresses AI agents, meaning programs that carry out tasks on their own. Gigabyte says the machine can keep long-running agents online all day while data and reasoning stay on the device. It also mentions Nvidia PAIR, short for Personal AI Router, which can spread inference work (running a trained model) across devices on a home or studio network.

For creative work, it points to ComfyUI for generative AI workflows. It also points to Nvidia Broadcast and RTX Video for improving audio and video. For games, it cites DLSS 5, which uses what Nvidia calls neural rendering to improve lighting and textures in supported games.

How to read a press release like this

This is a company announcement, so it describes the product in the best light. It makes no performance claims you can check, such as speeds for particular models, and it gives no price. The pitch rests on a few ideas: privacy, responsiveness and cost control.

Those ideas are real trade-offs. Local AI can keep your files off third-party servers and avoid per-use fees. But you pay for the hardware up front, and you are limited to models that fit in the machine's memory. Whether this desktop is good value depends on numbers the release does not provide.

The release also says tools are "compatible" or "supported" and that features work in "supported games". Those qualifiers mean not everything will run well, or at all, so check the exact software you care about.

Who this is for

The target buyers are developers, creators and heavy AI users who want a capable machine at their desk. A small team testing AI agents, or a video creator who wants AI tools without cloud bills, fits that description. A casual user who occasionally asks a chatbot questions does not.

For everyone else, the useful takeaway is the trend. Hardware makers are now selling local AI as a regular desktop feature, with a Windows environment and familiar tools. That gives people a real alternative to renting computing power by the task.

What to watch

Try it

Before considering any local-AI machine, add up what you spend on cloud AI in a year, then compare that with the hardware price once Gigabyte publishes it for your region.

Sources and links

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 Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by PR Newswire · 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.

More AI news