DeepSeek's New LLM Engine Just Made GPUs 3x More Valuable
From bycloud · · 19 min

The player loads from YouTube only when you click. It uses YouTube's privacy-enhanced mode.
If the video does not play, the owner may have turned off embedding. Watch it on YouTube instead.
Description
Benchmark and optimize your AI coding agents now with Warp Factories http://warp.dev/bycloud
DeepSeek just open-sourced DeepInfer, its highly optimized inference engine capable of delivering over 26,000 output tokens per second per GPU. In this video, I'll be breaking down and explore how DeepSeek is changing the economics of AI inference.
my latest projects: Your private local AI server needs an upgrade or maintenance support? Book a free audit now at https://nimbuskernel.com/
Intuitive AI Academy - Learn LLMs Intuitively We just upgrade our entire fundamental section! https://intuitiveai.academy/ limited time code "BACK2SCHOOL" for 50% off yearly plan
My Newsletter (weekly top research papers) https://mail.bycloud.ai/
My Patreon https://www.patreon.com/c/bycloud
Sources: DeepInfer https://github.com/deepseek-ai/DeepInfer DeepSeek Open Source Week https://github.com/deepseek-ai/open-infra-index
Try out my new fav place to learn how to code https://scrimba.com/?via=bycloudAI
This video is supported by the kind Patrons & YouTube Members: 🙏Spam Maj, Alex, Chris LeDoux, DX Research Group, Poof N' Inu, Deagan, Robert Zawiasa, Ryszard Warzocha, Tobe2d, Louis Muk, Akkusativ, Kevin Tai, Mark Buckler, NO U, Tony Jimenez, Ângelo Fonseca, jiye, Anushka, Asad Dhamani, Binnie Yiu, Calvin Yan, Clayton Ford, Diego Silva, Etrotta, Gonzalo Fidalgo, Handenon, Hector, Jake Disco very, Michael Brenner, Nilly K, OlegWock, Daddy Wen, Shuhong Chen, Sid_Cipher, Stefan Lorenz, Sup, tantan assawade, Thipok Tham, Thomas Di Martino, Thomas Lin, Richárd Nagyfi, Paperboy, mika, Leo, Berhane-Meskel, Kadhai Pesalam, mayssam, Bill Mangrum, nyaa, Toru Mon, Lame Plane, Matej Macak, Len Mo, saylikhapekar, ZyanSheep, THEVIERAOS, Ricardo Raphael Corona-Moreno, superchordate
[Discord] https://discord.gg/NhJZGtH [Twitter] https://twitter.com/bycloudai [Patreon] https://www.patreon.com/bycloud [Business Inquiries] bycloud@smoothmedia.co [Other Inquiries] bycloudai@gmail.com [Profile & Banner Art] https://twitter.com/pygm7 [Video Editor] @Booga04 Manim Animations created by https://x.com/miyashita_03 [Ko-fi] https://ko-fi.com/bycloudai [Bitcoin (BTC)] 3JFMJQVGXNA2HJE5V9qCwLiqy6wHY9Vhdx [Ethereum (ETH)] 0x3d784F55E0bE5f35c1566B2E014598C0f354f190 [Litecoin (LTC)] MGHnqALjyU2W6NuJSSW9fTWV4dcHfwHZd7 [Bitcoin Cash (BCH)] 1LkyGfzHxnSfqMF8tN7ZGDwUTyBB6vcii9 [Solana (SOL)] 6XyMCEdVhtxJQRjMKgUJaySL8cGoBPzzA2NPDMPfVkKN
0:00 Intro 2:40 How DeepSeek Makes $38 Per GPU-Hour 3:50 Why LLM Inference Is So Inefficient 4:23 Prefill vs. Decoding 6:02 Throughput vs. Latency 9:47 DeepSeek's 96% KV Cache Hit Rate 11:25 The Problem with This Engine 13:17 How much is DeepSeek really making? 16:31 Conclusion