<?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 h100</title>
    <link>https://minutesof.com/dylan-patel/on/h100/</link>
    <description>Everything Dylan Patel has said on h100: 9 verbatim quotes between August 2023 and April 2026, each with a timestamp and a link to the recording it came…</description>
    <language>en</language>
    <lastBuildDate>Sun, 30 Aug 2026 16:22:17 +0000</lastBuildDate>
    <atom:link href="https://minutesof.com/dylan-patel/on/h100/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Patel reports H100 rental prices increased from $1.71 per hour to over $2.40 in six months.</title>
      <link>https://minutesof.com/q/525bb53d-bbc6-4eaf-84ff-3b55976574fc/</link>
      <guid isPermaLink="true">https://minutesof.com/q/525bb53d-bbc6-4eaf-84ff-3b55976574fc/</guid>
      <description>“And just in the last six months, it&#x27;s gone from, you know, deals for one year transacting at $171.60 an hour for h 1 hundreds to now $2.40 plus.” — Aria Networks</description>
      <pubDate>Thu, 16 Apr 2026 02:27:54 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel reports seeing AI labs sign GPU rental deals at $2.40 per hour for H100s, well above initial $1.40 cost.</title>
      <link>https://minutesof.com/q/4203d5a5-7268-4115-a39f-6b148314d682/</link>
      <guid isPermaLink="true">https://minutesof.com/q/4203d5a5-7268-4115-a39f-6b148314d682/</guid>
      <description>“I&#x27;ve seen deals where certain AI labs I&#x27;m be a little bit vague here for a reason, have signed at as high as $2.40 for two to three years.” — Dwarkesh Patel</description>
      <pubDate>Fri, 13 Mar 2026 16:26:57 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel says H100 cannot physically reach advertised 2,000 teraflops, maxing at 1,300 teraflops for matrix opera</title>
      <link>https://minutesof.com/q/7eb3f152-085c-4b25-9dd3-33f9d606c0a6/</link>
      <guid isPermaLink="true">https://minutesof.com/q/7eb3f152-085c-4b25-9dd3-33f9d606c0a6/</guid>
      <description>“In fact, it&#x27;s physically impossible to get 2,000 teraflops out of the H100. Even though the advertises it, at most you can get 1,300 FP8 floating point eight teraflops.” — Clockwork</description>
      <pubDate>Fri, 21 Nov 2025 17:18:51 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel finds 10x performance TCO advantage for Blackwell over H100 at certain operating points.</title>
      <link>https://minutesof.com/q/7eaa498f-3259-4e8c-a196-98d19ea3ec72/</link>
      <guid isPermaLink="true">https://minutesof.com/q/7eaa498f-3259-4e8c-a196-98d19ea3ec72/</guid>
      <description>“But at other points, you can have a 10x performance TCO benefit from going with a more expensive Blackwell server than a cheaper H100 server.” — Open Compute Project</description>
      <pubDate>Thu, 23 Oct 2025 06:01:19 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel notes CoreWeave got H100 services online six months before every hyperscaler in many cases.</title>
      <link>https://minutesof.com/q/c2e290f0-d29d-46af-b1bb-46bc871b30f2/</link>
      <guid isPermaLink="true">https://minutesof.com/q/c2e290f0-d29d-46af-b1bb-46bc871b30f2/</guid>
      <description>“Get services online with h one hundreds before every hyperscaler by a factor of, like, as much as, like, six months in many cases.” — CoreWeave</description>
      <pubDate>Fri, 23 May 2025 17:27:49 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel says NVIDIA will sell over 3 million total GPUs next year and over a million H100s this year.</title>
      <link>https://minutesof.com/q/2855439b-9e6d-4778-82dc-a589fdbb9562/</link>
      <guid isPermaLink="true">https://minutesof.com/q/2855439b-9e6d-4778-82dc-a589fdbb9562/</guid>
      <description>“NVIDIA is gonna sell well over 3,000,000 total GPUs next year, over a million H100s this year alone. There&#x27;s a lot of GPU capacity coming online. It&#x27;s an incredible amount.” — Latent Space</description>
      <pubDate>Tue, 05 Dec 2023 18:16:28 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel states NVIDIA will sell over one million H100s this year and over three million total GPUs next year.</title>
      <link>https://minutesof.com/q/a5596c89-b7bc-42e6-8a13-449244f817e4/</link>
      <guid isPermaLink="true">https://minutesof.com/q/a5596c89-b7bc-42e6-8a13-449244f817e4/</guid>
      <description>“NVIDIA is gonna sell well over 3,000,000 total GPUs next year, over a million H100s this year alone. There&#x27;s a lot of GPU capacity coming online.” — Latent Space</description>
      <pubDate>Tue, 05 Dec 2023 18:16:28 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel forecasts NVIDIA will ship over one million H100 and A100 GPUs this year.</title>
      <link>https://minutesof.com/q/7ad7302b-f526-4151-806b-845410bbaa88/</link>
      <guid isPermaLink="true">https://minutesof.com/q/7ad7302b-f526-4151-806b-845410bbaa88/</guid>
      <description>“NVIDIA is gonna ship over a million, h one hundreds plus a a one hundreds this year.” — The Inside View</description>
      <pubDate>Wed, 09 Aug 2023 15:00:23 +0000</pubDate>
      <category>h100</category>
    </item>
    <item>
      <title>Patel states NVIDIA will build and sell approximately 400,000 H100s in Q3.</title>
      <link>https://minutesof.com/q/62d0ab02-d94f-4006-9057-aec70b68289f/</link>
      <guid isPermaLink="true">https://minutesof.com/q/62d0ab02-d94f-4006-9057-aec70b68289f/</guid>
      <description>“they&#x27;re gonna build at about 400,000 GPUs and and sell about 400,000 h one hundreds in q three.” — The Inside View</description>
      <pubDate>Wed, 09 Aug 2023 15:00:23 +0000</pubDate>
      <category>h100</category>
    </item>
  </channel>
</rss>
