clm_41df33632f644ae9b1cbfed7cecf8ba0_r1

2025-06-13_amd-advancing-ai-mi350x-and-mi400-ualoe72-mi500-ual256::AMD25-0001

① SA Source

• Source: 開啟完整 SA 文章
• Section: AMD Advancing AI: MI350X and MI400 UALoE72, MI500 UAL256
• Line hint: 17

Context Before

---

id: 174558644
title: “AMD Advancing AI: MI350X and MI400 UALoE72, MI500 UAL256”
subtitle: “Software Improvement, Marketing RDFs, AMD Fostering Neocloud, MI355 is not Rack Scale, MI400 is UALoE, Not UALink”
published_at: “2025-06-13T15:38:21.000Z”
authors: [“Kimbo Chen”, “Dylan Patel”, “Daniel Nishball”, “Wega Chu”, “Ivan Chiam”, “Gerald Wong”, “Patrick Zhou”]
url: “https://newsletter.semianalysis.com/p/amd-advancing-ai-mi350x-and-mi400-ualoe72-mi500-ual256”
audience: “only_paid”
access: authenticated_subscription
assets: local-v2
source_selector: “.available-content”
collected_at: “2026-08-08T13:34:24.629331+00:00”

AMD Advancing AI: MI350X and MI400 UALoE72, MI500 UAL256

Evidence

At its Advancing AI 2025 event, AMD launched the MI350X/MI355X GPUs which could be competitive to Nvidia’s HGX B200 solutions for inference of small to medium LLMs on a performance per TCO basis.

Context After

Instead, it is the MI400 Series that is a true rack scale solution that could potentially be competitive with Nvidia’s VR200 NVL144 rack scale solutions in H2 2026. There is also some marketing spin around the MI400 Series as AMD has renamed its “IF over Ethernet” protocol to “UALink Protocol over Ethernet”, which is not real UALink.

In this article, we will discuss the relative competitiveness of AMD’s new products and analyze their total cost of ownership. We will also elaborate on AMD’s new hyperscale customer, AWS, and on the flip side, the continued disappointment in follow-on orders on from existing customer Microsoft.

② Atomic Claim

AMD 在 Advancing AI 2025 推出 MI350X/MI355X；SemiAnalysis 認為其在 small-to-medium LLM inference 的 perf/TCO 有機會與 HGX B200 競爭。

• Epistemic Mode: INFERRED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "frame_type": "NARY_RELATION",
  "participants": [
    {
      "node": {
        "id": "02_companies/AMD",
        "label": "AMD"
      },
      "role": "vendor"
    },
    {
      "node": {
        "id": "04_knowledge_base/MI350X",
        "label": "MI350X"
      },
      "role": "gpu"
    },
    {
      "node": {
        "id": "04_knowledge_base/MI355X",
        "label": "MI355X"
      },
      "role": "gpu"
    },
    {
      "node": {
        "id": "04_knowledge_base/NVIDIA B200",
        "label": "B200"
      },
      "role": "competitor"
    },
    {
      "node": {
        "id": "04_knowledge_base/Large language model",
        "label": "LLM"
      },
      "role": "workload"
    }
  ],
  "qualifiers": {
    "condition_text": "small-to-medium LLM inference perf/TCO",
    "numeric_mentions": [
      "2025"
    ],
    "temporal_mentions": []
  },
  "relation_type": "PRODUCT_LAUNCH_AND_POSITIONING"
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
vendor	AMD	AMD
gpu	MI350X	MI350X
gpu	MI355X	MI355X
competitor	B200	04_knowledge_base/NVIDIA B200
workload	LLM	04_knowledge_base/Large language model

⑤ Human Review

請在 Properties 逐項確認：

• 原文 → Atomic Claim 是否忠實
• Atomic Claim → Semantic Frame 是否忠實
• Canonical Entity mapping 是否正確
• Epistemic mode 是否保留原文語氣
• 最後選擇 review_action

Review state

Markdown 內文不是正式 approval。只有 Apply bridge 寫入的 Decision Ledger event 才是正式決策。