<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>trainium: everyone on the record — The Minutes</title>
    <link>https://minutesof.com/on/trainium/</link>
    <description>Everyone on the record about trainium: 7 verbatim quotes from 3 people, December 2024 to June 2026, in one chronology, each with a timestamp and a link to…</description>
    <language>en</language>
    <lastBuildDate>Sun, 30 Aug 2026 22:33:18 +0000</lastBuildDate>
    <atom:link href="https://minutesof.com/on/trainium/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Dylan Patel: Patel reports Trainium rents for under $10 billion per gigawatt while GPUs cost $12-13 billion.</title>
      <link>https://minutesof.com/q/32312d2c-7043-4e82-9029-8dcbaeafc456/</link>
      <guid isPermaLink="true">https://minutesof.com/q/32312d2c-7043-4e82-9029-8dcbaeafc456/</guid>
      <description>“Tranium sells at sub $10,000,000,000 per gigawatt rental rate to Anthropic and to OpenAI. GPUs, at least before the craziness of the last six months, usually went around 12 to $13,000,000,000 per gigawatt.” — Dylan Patel, Sequoia Capital</description>
      <pubDate>Tue, 30 Jun 2026 12:00:25 +0000</pubDate>
      <category>trainium</category>
    </item>
    <item>
      <title>Gavin Baker: Baker states only NVIDIA and Amazon Trainium have functioning switched scale-up networks for infe</title>
      <link>https://minutesof.com/q/05badf52-27d8-4742-8532-91f442850ac7/</link>
      <guid isPermaLink="true">https://minutesof.com/q/05badf52-27d8-4742-8532-91f442850ac7/</guid>
      <description>“And the only two functioning switched scale up networks in the world today are the ones that power NVIDIA GPUs and Amazon&#x27;s Trainiums.” — Gavin Baker, Sohn Conference Foundation</description>
      <pubDate>Fri, 15 May 2026 18:17:41 +0000</pubDate>
      <category>trainium</category>
    </item>
    <item>
      <title>Gavin Baker: Baker argues Trainium is most underestimated because frontier mixture-of-expert models require sw</title>
      <link>https://minutesof.com/q/9f8eb50e-f98e-49c0-8051-f5a028e01a39/</link>
      <guid isPermaLink="true">https://minutesof.com/q/9f8eb50e-f98e-49c0-8051-f5a028e01a39/</guid>
      <description>“And so Trainium is for sure the most underestimated, not only because of those design choices, but because the all of these frontier models are what are called mixture of expert models.” — Gavin Baker, Sohn Conference Foundation</description>
      <pubDate>Fri, 15 May 2026 18:17:41 +0000</pubDate>
      <category>trainium</category>
    </item>
    <item>
      <title>Gavin Baker: Tranium, by far. Tranium is going to be to 2026, especially in the second half of this year when</title>
      <link>https://minutesof.com/q/a91f962e-6711-454f-b1db-5c6d4437360d/</link>
      <guid isPermaLink="true">https://minutesof.com/q/a91f962e-6711-454f-b1db-5c6d4437360d/</guid>
      <description>“Tranium, by far. Tranium is going to be to 2026, especially in the second half of this year when Tranium three really ramps, as TPUs were to twenty twenty five.” — Gavin Baker, Sohn Conference Foundation</description>
      <pubDate>Fri, 15 May 2026 18:17:41 +0000</pubDate>
      <category>trainium</category>
    </item>
    <item>
      <title>Bill Gurley: Patel reveals Amazon and Anthropic are building a 400,000 Trainium chip supercomputer as evidence</title>
      <link>https://minutesof.com/q/f82e7a79-273d-4b6e-8272-b34b89b6482f/</link>
      <guid isPermaLink="true">https://minutesof.com/q/f82e7a79-273d-4b6e-8272-b34b89b6482f/</guid>
      <description>“Amazon and Anthropic have decided to, you know, make a 400,000 Trainium server supercomputer. Right? 400,000 chips. Right? Going back to the whole scaling law is dead.” — Bill Gurley, Bg2 Pod</description>
      <pubDate>Mon, 23 Dec 2024 20:25:20 +0000</pubDate>
      <category>trainium</category>
    </item>
    <item>
      <title>Bill Gurley: Patel says Amazon and Anthropic building 400,000 chip Trainium supercomputer proves they believe</title>
      <link>https://minutesof.com/q/fb579459-b9f0-48bb-b811-bb4cfc02123b/</link>
      <guid isPermaLink="true">https://minutesof.com/q/fb579459-b9f0-48bb-b811-bb4cfc02123b/</guid>
      <description>“And and because of this, right, Amazon and Anthropic have decided to, you know, make a 400,000 Trainium server supercomputer. Right? 400,000 chips. Right? Going back to the whole scaling law is dead.” — Bill Gurley, Bg2 Pod</description>
      <pubDate>Mon, 23 Dec 2024 20:25:20 +0000</pubDate>
      <category>trainium</category>
    </item>
    <item>
      <title>Dylan Patel: Patel dubs Amazon&#x27;s Trainium chip the Amazon Basics TPU due to less efficient but cost-effective</title>
      <link>https://minutesof.com/q/55fd7d23-c6ff-42d1-ada9-ae49a012dcd3/</link>
      <guid isPermaLink="true">https://minutesof.com/q/55fd7d23-c6ff-42d1-ada9-ae49a012dcd3/</guid>
      <description>“Yeah. So so funnily enough, Amazon&#x27;s chip is the Amazon I I call it the Amazon&#x27;s basics TPU. Right?” — Dylan Patel, BG2 Pod</description>
      <pubDate>Mon, 23 Dec 2024 20:12:56 +0000</pubDate>
      <category>trainium</category>
    </item>
  </channel>
</rss>
