<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>The Minutes of Dylan Patel on failures</title>
    <link>https://minutesof.com/dylan-patel/on/failures/</link>
    <description>Everything Dylan Patel has said on failures: 2 verbatim quotes between March 2025 and November 2025, each with a timestamp and a link to the recording it…</description>
    <language>en</language>
    <lastBuildDate>Sun, 30 Aug 2026 16:22:23 +0000</lastBuildDate>
    <atom:link href="https://minutesof.com/dylan-patel/on/failures/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Patel reports five or six neo clouds have already gone out of business in just a few years.</title>
      <link>https://minutesof.com/q/aa801cea-f3ca-4fe8-aabc-405fb076edbe/</link>
      <guid isPermaLink="true">https://minutesof.com/q/aa801cea-f3ca-4fe8-aabc-405fb076edbe/</guid>
      <description>“There&#x27;s I believe five or six that have already gone out of business. It&#x27;s only been a few years. Right. Right?” — Clockwork</description>
      <pubDate>Fri, 21 Nov 2025 17:18:51 +0000</pubDate>
      <category>failures</category>
    </item>
    <item>
      <title>Patel reports Meta&#x27;s 16,000 GPU cluster had failures every few hours while next-generation models train on 100</title>
      <link>https://minutesof.com/q/04dfd209-35a0-4bd2-95b6-e803801d15c4/</link>
      <guid isPermaLink="true">https://minutesof.com/q/04dfd209-35a0-4bd2-95b6-e803801d15c4/</guid>
      <description>“And that was a measly 16,000 GPUs. Right? And when you look at like the next generation Llamas, the next generation GPTs, etcetera, they&#x27;re training on a 100,000 GPUs or more.” — Prime Intellect AI</description>
      <pubDate>Tue, 25 Mar 2025 18:21:45 +0000</pubDate>
      <category>failures</category>
    </item>
  </channel>
</rss>
