B-TIER news
AMD Ryzen AI Max+ 395 — 128 GB unified memory for local 200B models
100k ft — What is this?
AMD Ryzen AI Max+ 395 — 128 GB unified memory for local 200B models
50k ft — Why care?
Lisa Su demos a palm-sized Ryzen AI Halo dev box on the Ryzen AI Max+ 395. Its 128 GB of unified memory is shared by CPU, GPU and NPU, enough to run models up to about 200B parameters offline. The caption makes the case for owning compute instead of renting it: private documents, no per-request costs, no usage caps.
10k ft — Key takeaways
• Local boxes trade convenience for privacy and no per-token costs.
• Memory capacity, not raw speed, limits which models you can run locally.
5k ft — Tools & techniques
AMD Ryzen AI Max+ 395
local inference, unified memory
1k ft — Deep-dive / try it
AMD CEO Lisa Su recently demonstrated a compact AI PC powered by AMD’s Ryzen AI Max+ 395, but the most important part was not raw speed. It was memory.
Ryzen AI Halo systems can support up to 128GB of unified memory, allowing the CPU and GPU to share one large memory pool. That gives demanding AI models far more room to run locally than they would have on most consumer computers.
This matters because large models need more than powerful processors. They also need enough memory to load, operate, and handle complex tasks without relying on cloud servers.
Until now, many developers, researchers, and founders have depended on rented GPUs, API credits, hosted tools, and expensive monthly subscriptions. More capable local hardware could start changing that equation.
Instead of paying continu…
Transcript excerpt
Now, I would say this is pretty beautiful. Do you guys agree? So let me tell you what it is. This is the smallest AI development system in the world, capable of running models with up to 200 billion parameters locally, not connected to anything. It's powered by our highest end Ryzen AI Max processor with 128 gigabytes of high-speed unified memory that is shared by CPU, GPU, and NPU. This architecture accelerates system performance and makes it possible to efficiently run large AI models on a compact desktop PC that fits in your hand.