<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>AI Infrastructure on AI Notes</title>
    <link>https://ai.ngchi.online/tags/ai-infrastructure/</link>
    <description>Recent content in AI Infrastructure on AI Notes</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Tue, 27 Oct 2026 09:19:14 +0700</lastBuildDate>
    <atom:link href="https://ai.ngchi.online/tags/ai-infrastructure/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>The Exposure Window Appears in Only One of the Two Reports</title>
      <link>https://ai.ngchi.online/posts/the-exposure-window-appears-in-only-one-of-the-two-reports/</link>
      <pubDate>Tue, 27 Oct 2026 09:19:14 +0700</pubDate>
      <guid>https://ai.ngchi.online/posts/the-exposure-window-appears-in-only-one-of-the-two-reports/</guid>
      <description>Cloudflare and the SGLang maintainers published coordinated reports on the same cross-request KV cache leak. Both are candid; the blast radius is in one and the twenty-month exposure window is in the other.</description>
    </item>
    <item>
      <title>Cerebras Published the Arithmetic. One Assumption Halves It</title>
      <link>https://ai.ngchi.online/posts/cerebras-published-the-arithmetic-one-assumption-halves-it/</link>
      <pubDate>Mon, 26 Oct 2026 09:16:58 +0700</pubDate>
      <guid>https://ai.ngchi.online/posts/cerebras-published-the-arithmetic-one-assumption-halves-it/</guid>
      <description>Cerebras published every step of fitting a 3T-parameter model onto 59 wafer-scale chips, ending at 190 concurrent requests. The architecture checks out; one disclosed cache assumption, which its own docs contradict, halves the answer.</description>
    </item>
  </channel>
</rss>
