Nvidia: The Machine Learning Company (2006-2022)
ฝัง
- เผยแพร่เมื่อ 21 พ.ย. 2024
- By 2012, NVIDIA was on a decade-long road to nowhere. Or so most rational observers of the company thought. CEO Jensen Huang was plowing all the cash from the company’s gaming business into building a highly speculative platform with few clear use cases and no obviously large market opportunity. And then... a miracle happened. A miracle that led not only to Nvidia becoming the 8th largest market cap company in the world, but also nearly every internet and technology innovation that’s happened in the decade since. Machines learned how to learn. And they learned it... on Nvidia.
PSA: We’re doing an ARENA SHOW!! May 4th, 2022 in Seattle (Star Wars day). All proceeds go to charity. We’d love to see you there! events.pitchbo...
If you want more Acquired, you can follow our newly public LP Show feed here in the podcast player of your choice (including Spotify!): pod.link/acqui...
Sponsors:
Thank you to our presenting sponsor for all of Season 10, Vanta! Vanta is the leader in automated security compliance - making SOC 2, HIPAA, GDPR, and more a breeze for startups and organizations of all sizes. You might say they’re like the “AWS of security and compliance”. Everyone in the Acquired community can get 10% off using this link: bit.ly/acquire...
Thank you as well to Vouch and to SoftBank Latin America!
bit.ly/acquire...
bit.ly/acquire...
Links:
Ben Thompson’s great Stratechery interview with Jensen: stratechery.co...
Linus Tech Tips tests an Nvidia A100: • NVIDIA REFUSED To Send...
Episode sources: docs.google.co...
Carve Outs:
The Expanse short story collection, Memory's Legion: www.amazon.com...
Sony RX100 point-and-shoot camera: electronics.so...
Note: Acquired hosts and guests may hold assets discussed in this episode. This podcast is not investment advice, and is intended for informational and entertainment purposes only. You should do your own research and make your own independent decisions when considering any financial transactions.
![BOWKYLION - วิงวอน (ex-change) [Official MV]](http://i.ytimg.com/vi/cLdnZWDaO-w/mqdefault.jpg)
0:00 intro
4:33 sponsor #1
7:47 recap
10:24 in-house drivers
15:07 turning to supercomputing
21:33 CUDA bet
23:33 AMD acquires ATI, stock drop
27:04 CUDA paradigm
31:06 proprietary
35:13 misadventures
35:52 Tegra, Tesla S, Switch
38:17 other GPUs to mobile
39:28 Android ecosystem
40:20 Acera & Graphcore
41:40 AI miracle
47:48 CUDA implementation
51:28 ad targeting, scale
53:58 stock up
57:58 matrix transforms
59:15 future?
1:00:58 crypto mining
1:05:39 data centers
1:10:58 connectivity
1:13:43 sponsor #2
1:13:47 ARM
1:21:04 valuation
1:25:58 licensing, segmentation
1:28:55 growth
1:30:02 ray-tracing, deep learning super-sampling
1:34:37 board partners, tuning
1:36:32 autonomous automotive
1:39:55 omniverse
1:42:27 bear & bull case, competition
1:49:38 advantage
1:52:43 hack, leak
1:54:44 picks & shovels
1:55:30 expand mission
1:56:33 not dying,
1:57:08 feature size
1:59:08 capex graph
2:01:31 sponsor #3
2:04:01 grading & prospects
2:10:19 outro
Thanks! It was so convenient for a 2 hours long videos
Q
wow this is great thanks so much Bruno!
Thanks!
I rarely type in comments but I’ve watched a few of these episodes now and honestly they’re so enjoyable, can’t believe you’re not a huge. Thanks for the great vids
Thank you! We have many more subscribers to the podcast feed right now vs TH-cam, but we're investing in growing YT now too!
I believe the Stanford quantum chemistry Professor was Todd Martínez who worked with Vijay Pande (now GP at Andreessen Horowitz) and that GPU project would turn into Folding@Home which would really take off when they started hosting the code on the PS3.
This is just such a fantastic episode and I’m very impressed by your breadth of knowledge including historical tech knowledge in this episode. Thank you for this great series on Nvidia.
high-quality conversation, 2 hours well spent; thanks!
Great Nvidia series !! You made my work time a pleasure. 😁 wish I invested back in 2017. Thank guys.
Watching in June of 2024 while Nvidias market cap is at 3 Trillion makes the evaluation section very interesting.
That Intro about modelling fire is great. Going down to elementary particle/fields accuracy in calculations will be possible for bigger and bigger systems. Implies how demand for compute will always scale up with costs coming down. I worked with models and physics, and using 1000x compute for more accurate result can often be achieved by a single shifting of the decimal place in the code. Who knows what physical powers better modelling will unlock.
Not enough comments nor likes for this quality video. I’m a huge Nvidia believer. I think the company is the next Apple and that Jensen is the best CEO out there.
1:34:35 Excellent podcast, thank you, except for one GLARING ERROR at 1:34:35-You've already praised Nvidia for their margins. Nvidia produces the single highest margin hardware component in any gaming rig; if they start competing on cards their costs would increase and their margins would decrease the more cards they sold. You estimated their IC gross margins at over 60%; Asus gross margins are a fraction of that. Can't imagine Jensen investing in a low-margin, highly competitive segment like that.
Excellent podcast!
Fantastic series on NVIDIA, thanks!
I am Jinhui from OneFlow. I watched your podcast and was inspired very much. I wonder whether we can translate your podcast into Chinese and publish it at our blog. I believe some others like me will learn a lot from your podcast and may come to watch the video if they have further interest. Thank you very much.
Best,
Jinhui
Из-за тебя здесь играю уже давно) пока выхожу в плюс и все норм получается)))
Best content on the net!!
great insight on their company. the share price is very up since this video
The conversation is always incredibly insightful.
I am incredibly interested in the research you mentioned in relation to speech recognition, machine learning models, and the grains of sand on the earth.
Is it possible for you to point me in the direction of the source?
Thank you for the insights.
Great episode!
1:00:30 There are definitely more. Computer for the last half century was used for primarily for communication, that is why CPU was built like this. Computer should be used for computation, which is embarrassing to be done in sequence. Deep learning is just one of thousands use cases requires massive parallel computing and I doubt it is the largest. You will find very soon the app after app after app to be accelerated as I said, computer is for computation.
Great episode
can you guys do a similar episode on AMD as well?
Small correction:
It’s Icera not acera
Google = Failure of Confidence. Basically a top notch research institute
1:19:45 : Well , excuse me but what also happened with the nVIDIA /ARM deal was basically this :
The Regulators forbid nVIDIA to do what in the past *was allowed* for AMD to do :
to be able to buy GPU I.P. when they were allowed to buy ATi. So while AMD was allowed this privilege ,to have(through purchase) both CPU I.P. and GPU I.P. under their control and to conduct their R&D , nVIDIA was NOT allowed through purchase to have both , and to me , *allowing someone to control both I.P. while you forbid another one to do the same , is unethical , immoral , and anti-competitive* .
As for ARM’s former CEO & founder going out back then and stating things such as : the U.K. interests will be damaged if this deal goes through and other similar comments , I would like to remind mr Hauser that ARM had *ALREADY been sold to a Japanese* company (obviously)outside of U.K. ,which theoretically could do whatever they wanted with their purchase regardless of U.K. interests… well ,anyway .. very nice video but “opened old wounds” lol !! 🙂
BUT NOT JENSEN.