“Dad, why didn’t you buy a ton of NVIDIA stock and become a millionaire? Didn’t you work on AI?”

My son hit me with that one recently. Ouch! He asked me the same question about Bitcoins last year. 藍

No, I didn’t retire by owning $NVDA. But I do own something that I chose: The knowledge of how GPUs actually work — threads, warps, blocks, CUDA.

Next week I am going to share that knowledge in my keynote lecture.

Title: GPU by hand ✍️ Together

Date: 10/30

Time: 10am (Pacific Time)

Register: https://luma.com/khuz5upe

The leap from Hopper to Blackwell isn’t just a story of “faster GPUs” — it’s a fundamental rethinking of how every layer of the stack is organized to unlock the next frontier of AI scale. And as typical in an AI by Hand lecture, instead of focusing on benchmarks from the outside, we will dive inside the black box — unpacking the math behind these advances and making the underlying mechanisms accessible to you.

Outline

Primer: Grid -> Block -> Warp -> Thread
Quantization: FP8 -> MXFP4
Scaling: Layer -> Micro-Tensor
Packaging: Single-Die -> Dual-Die
Communication: In-GPU -> In-switch reduction

In the end…

do I want to be remembered as a lucky dad,

or a devoted dad who helped him understand the world he’s growing into?

RSVP

This keynote lecture is sponsored by Together AI.

","contentEncoded","freeContentEncoded","textContent","“Dad, why didn’t you buy a ton of NVIDIA stock and become a millionaire? Didn’t you work on AI?”My son hit me with that one recently. Ouch! He asked me the same question about Bitcoins last year. 藍No, I didn’t retire by owning $NVDA. But I do own something that I chose: The knowledge of how GPUs actually work — threads, warps, blocks, CUDA.Next week I am going to share that knowledge in my keynote lecture.Title: GPU by hand ✍️ TogetherDate: 10/30Time: 10am (Pacific Time)Register: https://luma.com/khuz5upeThe leap from Hopper to Blackwell isn’t just a story of “faster GPUs” — it’s a fundamental rethinking of how every layer of the stack is organized to unlock the next frontier of AI scale. And as typical in an AI by Hand lecture, instead of focusing on benchmarks from the outside, we will dive inside the black box — unpacking the math behind these advances and making the underlying mechanisms accessible to you.OutlinePrimer: Grid -> Block -> Warp -> ThreadQuantization: FP8 -> MXFP4Scaling: Layer -> Micro-TensorPackaging: Single-Die -> Dual-DieCommunication: In-GPU -> In-switch reductionIn the end…do I want to be remembered as a lucky dad,or a devoted dad who helped him understand the world he’s growing into?RSVPThis keynote lecture is sponsored by Together AI.","published","strategy","content-ready","locked","deletedAt","isSyncedFromOwnBlog","original","duplicatesCount","duplicateVersion","originalArticleId","errorsCount","appearance","color","backgroundColor","shortSummary","summary","bullets","persons","places","words","sentiment","subject","meta","eventId","07851f4ca03f3573cc99cd6b7a28f2c15e60b851313e876879090f834ed4239c","nostrRootEvent","event",{"_18":187,"_191":192,"_193":194,"_71":195,"_207":25,"_157":208,"_209":210},"sig","421b0f1e56e509e45d043756c12cc25e14cd9fffc094dc0000fca7bcc2e006029de30aac5b06f0992f9e1103fc21803978959d153d28192f8268c2688e7bcfa9","kind",1,[196,197,198,202,205],[184,143],[34,149],[199,200,15,201],"a","30023:fa1756185f2879c4d47b882789bc588d3ad64030bf50c2f7e48eb57d2d8d02a8:db560c44-57ae-4d34-ba18-4da0a916b257","mention",[203,204],"client","Layer3.press",[206,25,15],"p","pubkey","GPU by Hand ✍️ Together\nKeynote Lecture Sponsored by Together AI\n\n\nnostr:naddr1qvzqqqr4gupzp7sh2cv972recn28hzp83x793rf66eqrp06sctm7fr4405kc6q4gqqjxgc34xccxxdp5956nwct9956xgve5943xzvfc956xgcfsvyunzdnzxg6nwa9yyas ","created_at",1761835769,"authorNPub","markdown","“Dad, why didn’t you buy a ton of NVIDIA stock and become a millionaire? Didn’t you work on AI?” \n \nMy son hit me with that one recently. Ouch! He asked me the same question about Bitcoins last year. 藍 \n \nNo, I didn’t retire by owning $NVDA. But I do own something that I chose: The knowledge of how GPUs actually work — threads, warps, blocks, CUDA. \n \nNext week I am going to share that knowledge in my keynote lecture.\n\n[\n\n](https://substackcdn.com/image/fetch/$s_!Upc8!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6602e63c-7d0d-489f-a7ee-9d37f40d87dd_3254x2148.jpeg)\n\n**Title:** GPU by hand ✍️ Together\n\n**Date:** 10/30\n\n**Time:** 10am (Pacific Time)\n\n**Register:** [https://luma.com/khuz5upe](https://luma.com/khuz5upe)\n\nThe leap from **Hopper** to **Blackwell** isn’t just a story of “faster GPUs” — it’s a fundamental rethinking of how every layer of the stack is organized to unlock the next frontier of AI scale. And as typical in an _AI by Hand_ lecture, instead of focusing on benchmarks from the outside, we will dive inside the black box — unpacking the math behind these advances and making the underlying mechanisms accessible to you.\n\n**Outline**\n\n* Primer: Grid -> Block -> Warp -> Thread\n \n* Quantization: FP8 -> MXFP4\n \n* Scaling: Layer -> Micro-Tensor\n \n* Packaging: Single-Die -> Dual-Die\n \n* Communication: In-GPU -> In-switch reduction\n \n\nIn the end… \n \ndo I want to be remembered as a lucky dad, \n \nor a devoted dad who helped him understand the world he’s growing into?\n\n[\n\n![](https://substackcdn.com/image/fetch/$s_!Ytt2!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F702526e9-c2d9-4433-ba64-8bb475253557_1320x614.png)\n\n](https://substackcdn.com/image/fetch/$s_!Ytt2!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F702526e9-c2d9-4433-ba64-8bb475253557_1320x614.png)\n\n[RSVP](https://luma.com/khuz5upe)\n\nThis keynote lecture is sponsored by [Together AI](https://www.together.ai/).","freeMarkdown","kind1EventId","kind30023EventId","6000701ec6f3b5ecd01c8c03b8eb04721a0c3c09d14a16fb7142d51b8a082b59","kind30023",{"_18":217,"_191":220,"_193":221,"_71":222,"_207":25,"_157":236,"_209":237},"9c40cc330c65f818538eaa7a4f084ee65efb778da6103fb1e14341aac27f04617751c7d19d8cde0726eab7fdf75e17feab4f3a14c9b2d549c736718d5f6cc906",30023,[223,226,227,230,231,233,234,235],[224,225],"r","https://fakepeter.syndichain.com/articles/db560c44-57ae-4d34-ba18-4da0a916b257",[30,143],[228,229],"published_at","1760988020",[34,149],[232,138],"d",[178,144],[203,204],[206,25,15],"GPU by Hand ✍️ Together\n“Dad, why didn’t you buy a ton of NVIDIA stock and become a millionaire? Didn’t you work on AI?” \n \nMy son hit me with that one recently. Ouch! He asked me the same question about Bitcoins last year. 藍 \n \nNo, I didn’t retire by owning $NVDA. But I do own something that I chose: The knowledge of how GPUs actually work — threads, warps, blocks, CUDA. \n \nNext week I am going to share that knowledge in my keynote lecture.\n\n[\n\n](https://substackcdn.com/image/fetch/$s_!Upc8!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6602e63c-7d0d-489f-a7ee-9d37f40d87dd_3254x2148.jpeg)\n\n**Title:** GPU by hand ✍️ Together\n\n**Date:** 10/30\n\n**Time:** 10am (Pacific Time)\n\n**Register:** [https://luma.com/khuz5upe](https://luma.com/khuz5upe)\n\nThe leap from **Hopper** to **Blackwell** isn’t just a story of “faster GPUs” — it’s a fundamental rethinking of how every layer of the stack is organized to unlock the next frontier of AI scale. And as typical in an _AI by Hand_ lecture, instead of focusing on benchmarks from the outside, we will dive inside the black box — unpacking the math behind these advances and making the underlying mechanisms accessible to you.\n\n**Outline**\n\n* Primer: Grid -> Block -> Warp -> Thread\n \n* Quantization: FP8 -> MXFP4\n \n* Scaling: Layer -> Micro-Tensor\n \n* Packaging: Single-Die -> Dual-Die\n \n* Communication: In-GPU -> In-switch reduction\n \n\nIn the end… \n \ndo I want to be remembered as a lucky dad, \n \nor a devoted dad who helped him understand the world he’s growing into?\n\n[\n\n![](https://substackcdn.com/image/fetch/$s_!Ytt2!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F702526e9-c2d9-4433-ba64-8bb475253557_1320x614.png)\n\n](https://substackcdn.com/image/fetch/$s_!Ytt2!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F702526e9-c2d9-4433-ba64-8bb475253557_1320x614.png)\n\n[RSVP](https://luma.com/khuz5upe)\n\nThis keynote lecture is sponsored by [Together AI](https://www.together.ai/).\nOpen original article:\nhttps://www.byhand.ai/p/gpu-by-hand-together fakepeter\n\n\n\nfakepeter",1761833657,"publishedToNostr","htmlContentKind30023","

“Dad, why didn’t you buy a ton of NVIDIA stock and become a millionaire? Didn’t you work on AI?”

My son hit me with that one recently. Ouch! He asked me the same question about Bitcoins last year. 藍

No, I didn’t retire by owning $NVDA. But I do own something that I chose: The knowledge of how GPUs actually work — threads, warps, blocks, CUDA.

Next week I am going to share that knowledge in my keynote lecture.

[

](https://substackcdn.com/image/fetch/$s_!Upc8!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6602e63c-7d0d-489f-a7ee-9d37f40d87dd_3254x2148.jpeg)

Title: GPU by hand ✍️ Together

Date: 10/30

Time: 10am (Pacific Time)

Register: https://luma.com/khuz5upe

Outline

\n
Primer: Grid -> Block -> Warp -> Thread
\n
\n
Quantization: FP8 -> MXFP4
\n
\n
Scaling: Layer -> Micro-Tensor
\n
\n
Packaging: Single-Die -> Dual-Die
\n
\n
Communication: In-GPU -> In-switch reduction
\n

In the end…

do I want to be remembered as a lucky dad,

or a devoted dad who helped him understand the world he’s growing into?

[

$\"\"$

](https://substackcdn.com/image/fetch/$s_!Ytt2!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F702526e9-c2d9-4433-ba64-8bb475253557_1320x614.png)

RSVP

This keynote lecture is sponsored by Together AI.

","pubDate","2025-10-20T19:20:20.000Z","modifiedAt","2025-10-30T14:14:17.450Z","storyId","likes","reposts","comments","tweets",[],{"_18":19,"_20":21,"_22":23,"_24":25,"_26":-5,"_27":28,"_29":15,"_30":31,"_32":33,"_34":35,"_36":15,"_37":38,"_39":40,"_41":15,"_42":43,"_44":45,"_46":45,"_47":48,"_49":252,"_82":254},{"_18":48,"_51":23,"_52":23,"_53":54,"_32":33,"_55":56,"_57":-5,"_58":-5,"_59":35,"_60":61,"_62":-5,"_63":-5,"_64":65,"_66":67,"_68":25,"_69":70,"_71":253,"_73":-5,"_74":75,"_76":61,"_77":61,"_78":-5,"_79":-5,"_80":-5,"_81":-5,"_44":45,"_46":45},[],{"_84":19,"_85":-5,"_86":-5,"_87":88,"_89":90,"_91":255,"_94":-5,"_95":-5,"_12":13,"_96":256,"_98":257},{"_93":75},[],{},"relatedArticles",[],"actionData","errors"]