There were a lot of computer graphics cards manufacturers back in the 90's. Nvidia survived amongst the whole pack of them. 3dfx, Matrox, S3, ATI, Cirrus. How much of that is soley due to Jenson vs outside factors outside of his control, only he really knows, but I think her deserves a bit more credit than right place right time. CUDA was a very expensive investment with zero guarantee of payoff. The market could have gone an entirely different way and Nvidia would have failed.
replyThat's fair. NVidia's execution has been rock solid for as long as they've been around and Huang deserves credit for that. But I would say most of the reason they've 100x'd since 2017 has been external forces.
reply> first crypto/blockchain, then AI, neither of which they really had any part in boosting.
replyNvidia boosts AI a lot. They are the central bank of AI. https://news.ycombinator.com/item?id=49673098
Maybe the verb tense was unclear. NVidia currently boosts AI, for obvious reasons. But OpenAI was the company who did all of the legwork to make AI into an industry, with some minor support from NVidia. NVidia wouldn't have 10x'd since 2022 without OpenAI.
replyAI has been a (minor but existing) industry before OpenAI. E.g. Deepmind exists since 2010, 6 years before OpenAIs founding. Sure, OpenAI has contributed to it, in great deals, but Nvidia has always been there during the growth of AI. In 2012, AlexNet has been trained on GTX 580 GPUs, in 2016 AlphaGo used Nvidia GPUs, etc.
replyYeah, they had been working on CUDA for a very long time before AI turned out to be interesting again.
replyTo be honest other hardware companies had years to catch up to Cuda and ML, but they invested shit into their software stacks.
replyEven when AI boom was imminient but 10 times smaller 2-3 years ago they did not invested heavily enough and Nvidia went all in.
I have no clue what percent of that decision making was Huang, but all the competition leadership was certainly incompetent.
So at least he did built company with a good leadership. Its a fact. Even if I personally dont like Nvidia or its products.
> they just happened to be in the right place, twice, in quick succession
replythat is generally what it looks like when someone is smarter than you
> that is generally what it looks like when someone is smarter than you
replyMaybe. This is also what everyone said about Elon for a long time, though.
It's not that Musk is not smart, he's just erratic, and growing increasingly so. He is the Howard Hughes of our day. They will find his body in an expensive hotel room at some point, surrounded by bottles of his own urine.
replyThe had the insight to develop and promote CUDA long before DL was a thing. IIRC AlexNet was when CUDA really paid off.
replyYeah, but in this case they also built the pickaxes and drills and did it better than any other pickaxe and drill builder. In your analogy, they are commodities and a shop like that can be set up by anyone. But in this case, that doesn't apply. And if you look at their sales composition, they never pivoted away from gaming, but data centers clearly became ludicrously more attractive in the AI era. But they were always very good at computing triangles, perhaps the best in the world. They were lucky that market forces made that computation very valuable but they were not lucky in their ability to build it.
replyI do agree! They execute extremely well and Huang deserves praise for that. I just don't see him as a visionary leader shaping the industry in the same way that e.g. Jobs was. In an alternate world where the NSA doesn't create bitcoin and OpenAI is never founded, NVidia would be a stable, reliable producer of consumer electronic parts.
replyhow can you be sure though we wont find a good use of computing triangles in that alternative universe? that technology would always have abstract value, how it is impmemented will depend on market forces but you have to admit, they know how to go with the flow, maybe to an extent where it seems like they just keep getting lucky
reply