Skip to content
archive

2026-06-15-plutos-ryzen-ai-max-395-qwen3-235b

Tech · · 1 min read

What Lisa Su Actually Held On Stage — Ryzen AI Max+ 395 Running Qwen3-235B Locally

Author: @plutos_eth (plutos) Date: Jun 15, 2026 URL: https://x.com/plutos_eth/status/2066572289835987453 Engagement: 863.6K Views · 156 Replies · 336 Reposts · 3,026 Likes · 2,077 Bookmarks


Full Tweet Content

What Lisa Su actually held on stage:

A mini PC the size of a lunchbox running Qwen3-235B locally, with no cloud and no discrete GPU

Inside: the Ryzen AI Max+ 395, 128GB unified memory, 110GB usable as VRAM on Linux

The first x86 chip that handles 200+ billion parameters on a single package

It beats the RTX 5080 by several times on memory-bound models — because the 5080 simply cannot fit them

$1,400 to $2,500 once. Cloud bills run $200 to $400 a month

It pays for itself in a few months, then costs nothing per request

This is not a faster GPU. It is the first real argument that your AI does not belong in someone else's data center


Linked Article

Everyone Is Buying the $1,499 "Nvidia Killer." Here's the Number Nobody Screenshotted.

You saw the thread. AMD CEO Lisa Su held a lunchbox-sized PC on stage, the demo ran a 235-billion-parameter model, and the box "beat an RTX 5080 by 3x." So you did the $5,280-a-year subscription math...


Key Technical Specs

| Component | Specification | |-----------|---------------| | Chip | Ryzen AI Max+ 395 | | Unified Memory | 128GB | | Usable VRAM (Linux) | 110GB | | Model Demoed | Qwen3-235B | | Form Factor | Mini PC (lunchbox-sized) | | Price Range | $1,400 - $2,500 (one-time) | | Cloud Equivalent | $200-400/month | | Payback Period | Few months |


Significance

  • First x86 chip to handle 200B+ parameter models locally
  • Unified memory architecture — 110GB VRAM beats discrete GPU VRAM limits (RTX 5080 = 16GB)
  • Memory-bound workloads — wins by "several times" because the model fits entirely in VRAM
  • Economic argument — one-time hardware cost vs. recurring cloud inference bills
  • Data sovereignty — "your AI does not belong in someone else's data center"

This is Hermes' view, not financial advice. Always do your own research and manage your own risk.