AMD releases powerful new mini-PC with 160 GB VRAM, 10 GbE and AMD Gorgon Halo

AMD has released a new mini-PC in the US, and it is aimed squarely at developers who want serious local AI horsepower in a compact box. The AMD Ryzen AI Halo Developer Platform, powered by the chip referred to as Gorgon Halo, arrives in the wake of recent Nvidia RTX Spark announcements. It aims to surpass its predecessor with 192 GB of RAM, of which 160 GB can be assigned as VRAM, along with 10 GbE networking.

Note: the details below come from the headline and description you provided. Where I go beyond them, I say so, and you should check specifics such as price, ports and exact chip specs against Notebookcheck's original report before publishing.

What AMD Has Actually Launched in the US

The new machine is called the AMD Ryzen AI Halo Developer Platform. It is a compact desktop system that AMD has now made available in the United States, positioning it as a tool for developers rather than a typical consumer mini-PC.

The launch follows a busy stretch in the mini-PC and AI hardware space. With Nvidia recently announcing its RTX Spark products, AMD appears keen to show that it has its own answer for people building and testing AI workloads locally.

The system is described as a successor to an earlier Halo-class machine. AMD's pitch is that this generation goes further on memory capacity and connectivity, two areas that matter most to anyone running large models.

Memory Capacity That Reaches 192 GB Total

The headline figure is the memory. The system carries 192 GB of RAM, a very large pool for a device of this size. Most mini-PCs top out far below that figure, which makes this configuration stand out.

Of that total, 160 GB can be assigned as VRAM. This unified memory approach lets the integrated graphics access a huge amount of memory, something that dedicated consumer graphics cards cannot match at similar price points.

For developers, that flexibility means the split between system memory and graphics memory can be tuned to the task. Heavy model workloads can claim most of the pool, while the rest stays available for the operating system and applications.

Why 160 GB of VRAM Matters for AI

Large language models and other generative AI systems are limited by memory more than almost anything else. A model that does not fit into VRAM either runs very slowly or cannot run at all on a given machine.

With 160 GB available to the GPU, developers can load much larger models locally, or run several at once. That opens the door to testing, fine-tuning and inference work without depending on cloud services or renting expensive data center GPUs.

Running locally also offers practical benefits such as data privacy, predictable costs and no network latency. For teams handling sensitive data, keeping everything on a desk-sized machine is a real draw.

Gorgon Halo and the Ryzen AI Silicon

The platform is built around AMD's Gorgon Halo processor, part of the company's Ryzen AI line. Halo parts combine CPU cores, a large integrated GPU and an NPU on one package, all sharing the same memory pool.

That design is what makes the large VRAM allocation possible. Rather than separate memory for the processor and graphics, everything draws from the same high-capacity pool.

I would suggest confirming the exact core counts, clock speeds and NPU performance figures from the source report, as the description provided does not list them in full.

10 GbE Networking for Faster Data Movement

Alongside the memory, the mini-PC includes 10 GbE networking. That is a notable inclusion in a compact system, since most mini-PCs ship with 1 GbE or 2.5 GbE at best.

Fast networking matters when moving large model files, datasets and checkpoints between machines. A multi-gigabyte model can take minutes to transfer on slower links, and 10 GbE cuts that time considerably.

It also makes the system a better fit for small clusters or lab setups. Developers can link multiple units or connect to network storage without the network becoming the bottleneck.

How It Compares With Its Predecessor

AMD says the new platform hopes to surpass the earlier Halo developer machine. The most visible gains are in memory capacity and networking, both aimed at the needs of AI developers.

A larger memory ceiling matters because model sizes keep growing. Hardware that was comfortable a year ago can feel cramped today, and AMD is clearly trying to stay ahead of that curve.

Specific benchmark comparisons were not included in the description, so real-world performance gains remain to be seen once independent reviews arrive.

The Competitive Pressure From Nvidia

The timing is hard to miss. Nvidia's RTX Spark announcements came shortly before this release, and both companies are now chasing the same audience of developers who want powerful local AI systems.

AMD's advantage here is its unified memory design and the sheer capacity on offer. Nvidia has the software ecosystem, particularly CUDA, which remains the default for many AI developers.

That software gap is the key question for AMD. Hardware specifications are impressive, but developer adoption will depend on how mature and easy to use AMD's AI software stack is on this platform.

What to Watch for Next

Pricing and availability details will be crucial. A machine with this much memory is unlikely to be cheap, and the cost will determine whether it reaches hobbyists and small teams or stays a niche tool.

Independent testing will also be important. Real-world results on model loading, token throughput, thermals and noise will show how well this compact design handles sustained loads.

For now, the AMD Ryzen AI Halo Developer Platform signals that high-memory local AI computing is becoming a serious battleground. If the final product delivers on its specifications, it could be one of the more interesting developer machines of the year.

If you have any questions, please don't hesitate to Contact Us