NVIDIA CEO Jensen Huang Reveals Breakthrough AI Chip at COMPUTEX 2024 (Supercut)

NVIDIA CEO Jensen Huang Reveals Breakthrough AI Chip at COMPUTEX 2024 (Supercut) Ticker Symbol: YOU 378K subscribers Join Subscribe 2.9K Share Download Thanks Clip Save 109,791 views Jun 2, 2024 #sora #chatgpt #nvda Highlights from the COMPUTEX 2024 Keynote by #nvidia CEO Jensen Huang. Topics include ‪@NVIDIA‬'s insane Blackwell B100 GPUs, the announcement of Blackwell Ultra in 2025, NVIDIA Rubin GPUs in 2026, NVIDIA Rubin Ultra GPUs in 2027, advanced AI networking technologies like NVIDIA's NVLink and Spectrum-X Ethernet, how they all come together to power generative AI technologies like #chatgpt and #sora by #openai and reshape computing as we know it. This is why I think Nvidia stock ( #nvda stock ) is still one of the best stocks to buy now. » Top Stocks for 2024: • Top 10 Stocks to Get Rich in 2024 (Wi... » Get My AI Stocks & Price Targets: https://www.tickersymbolyou.com/survey » Main Street Data (Best Financial Charts): https://mainstreetdata.com/subscripti... » Simply Wall Street's Nvidia (NVDA Stock) Valuation: https://simplywall.st/stocks/us/semic... Timestamps for this NVIDIA COMPUTEX 2024 Keynote Supercut: 00:00 Jensen Huang Announces NVIDIA Rubin GPU 02:20 NVIDIA Blackwell for AI Supercomputing 06:10 NVIDIA Networking: NVLink & Spectrum-X Ethernet 11:56 NVIDIA Robotics: Self-driving Cars & Humanoids » Podcast: • E5: Our Top 5 Stock Market & AI Predi... » Twitter: / tickersymbolyou » Patreon: / tickersymbolyou Resources & References: » NVIDIA CEO Jensen Huang Keynote at COMPUTEX 2024: • NVIDIA CEO Jensen Huang Keynote at CO... » Nvidia GPU Advances: • Nvidia's Breakthrough AI Chip Defies ... » Nvidia AI Advances: • Bigger Than ChatGPT?! Every NVIDIA GT... » OpenAI Dev Day: • OpenAI & Microsoft's HUGE AI Announce... » Google AI War: • The AI War Has Begun! Every Google I/... 🙏 Thanks for watching! Disclosures: All content on this channel is for informational purposes only and should not be construed as professional financial advice or recommendation to buy or sell any securities. Trading stocks, ETFs, other securities, and/or cryptocurrencies poses a considerable risk of loss. Neither host or guests can be held responsible for any direct or incidental loss incurred by applying any of the information offered. The content in this video is solely the opinions of the speaker who is not a licensed financial advisor or registered investment advisor. Should you need such advice, consult a licensed financial or tax advisor. When you make purchases through links in this video description, the author may earn a commission. See additional disclosures at https://tickersymbolyou.com/disclosures/ Chapters View all Transcript Follow along using the transcript. Show transcript Transcript Jensen Huang Announces NVIDIA Rubin GPU 0:00 this last 60 years we saw several 0:03 tectonic shifts in Computing where 0:07 everything changed and we're about to 0:09 see that happen again the further 0:11 performance we drive up the Greater the 0:13 cost decline Hopper platform of course 0:16 was the most successful data center 0:18 processor probably in history however 0:22 Blackwell is here and every single 0:25 platform as you'll notice are several 0:27 things you got the CPU you have the GPU 0:29 you have MV link you have to Nick and 0:30 you have to switch every single 0:33 generation as you'll see is not just a 0:35 GPU but it's an entire platform we build 0:38 the entire platform we integrate the 0:41 entire platform into an AI Factory 0:42 supercomputer however then we 0:44 disaggregate it and offer it to the 0:46 world our basic philosophy is very 0:48 simple one build the entire data center 0:51 scale dis agregate it and sell it to you 0:54 in Parts on a one-year Rhythm and we 0:57 push everything that technology limits 0:59 what whatever tsmc processed technology 1:03 will push it to the absolute limits 1:05 whatever packaging technology push it to 1:07 the absolute limits whatever memory 1:08 technology push it to Absolute limits 1:10 cies technology Optics technology 1:12 everything is pushed to the Limit while 1:15 Blackwell is here next year is Blackwell 1:18 Ultra just as we had h100 and h200 it'll 1:23 probably see some pretty exciting New 1:25 Generation from us for Blackwell Ultra 1:28 well this is the very first time 1:30 and I'm not sure yet whe I'm going to 1:32 regret this or not we have code names in 1:34 our company and we try to keep them very 1:37 secret most of the employees don't even 1:39 know but our next Generation platform is 1:41 called Reuben so we have the Reuben 1:43 platform and one year later we have the 1:45 Reuben um Ultra platform all of these 1:48 chips that I'm showing you here are all 1:50 in full development 100% of them and the 1:53 rhythm is one year at the limits of 1:56 Technology all 100% architecturally 1:58 compatible so this is 2:00 this is basically what Nvidia is 2:02 building and all of the riches of 2:04 software on top of it so in a lot of 2:06 ways the last 12 years the company has 2:09 really transformed tremendously and I 2:11 want to thank all of our partners here 2:12 for supporting us every step along the 2:15 way this is the Nvidia Blackwell 2:19 platform ladies and gentlemen this is NVIDIA Blackwell for AI Supercomputing 2:22 Blackwell this is our production board 2:26 this is the most complex highest 2:28 performance computer the world ever made 2:31 this is the gray 2:32 CPU and these are you could see each one 2:36 of these blackw dieses two of them 2:38 connected together you see that it is 2:41 the largest D the largest chip the world 2:43 makes and then we connect two of them 2:46 together with a 10 terte pers second 2:48 link and the performance is 2:51 incredible take a look at this the AI 2:54 flops uh for each generation has 2:57 increased by a, times in eight years and 3:01 so just to compare even Mo's law at its 3:05 best of 3:06 times compared to what Blackwell could 3:09 do so the amount of computations is 3:12 incredible and when whenever we bring 3:14 the computation High the thing that 3:17 happens is the cost goes down the amount 3:19 of energy that is used has gone down by 3:22 350 times Well Pascal would have taken 3:27 $1,000 3:29 gwatt hours 1,000 gwatt hours means that 3:33 it would take a gwatt data center the 3:36 world doesn't have a gwatt data center 3:38 but if you had a gigawatt data center it 3:40 would take a month if you had a 100 watt 3:44 100 megawatt data center it would take 3:46 about a year and that's the reason why 3:48 these large language models chat GPT 3:50 wasn't possible only eight years ago 3:53 with Blackwell what used to be 1,000 3:55 gwatt hours to three an incredible 3:58 advance our token generation performance 4:03 has made it possible for us to drive the 4:04 energy down by three four 45,000 times 4:08 okay so Blackwell is just an enormous 4:11 leap well even so it's not big enough 4:15 and so we have to build even larger 4:18 machines and so the way that we build it 4:20 is called dgx so this is a dgx Blackwell 4:25 this has this is air cooled has eight of 4:28 these gpus inside look at the size of 4:32 the heat sinks on these 4:36 gpus about 15 kilow 15,000 watts and 4:42 completely air cooled this version 4:44 supports x86 and it's it goes into the 4:48 infrastructure that we've been shipping 4:50 Hoppers into however if you would like 4:52 to have liquid cooling we have a new 4:54 system it's based on this board and we 4:57 call it mgx for modular so this one node 5:02 has four Blackwell chips and these 5:04 switches connect every single one of 5:07 these Blackwells to each other so that 5:10 we have one giant 72 GPU Blackwell this 5:15 now looks like one GPU this one GPU has 5:18 72 versus the last generation of eight 5:21 so we increased it by nine times the 5:23 amount of bandwidth we've increased by 5:25 18 times the AI flops we've increased by 5:28 45 times and yet the amount of power is 5:31 only 10 times this is 100 kilow and that 5:35 is 10 kilow and that's for one you could 5:39 always connect more of these together 5:40 and I'll show you how to do that in a 5:41 second there's this confusion about what 5:43 Nvidia does how is it 5:46 possible that that Nvidia became so big 5:49 building gpus and so there's an 5:52 impression that this is what a GPU looks 5:54 like now this is a GPU this is one of 5:57 the most advanced gpus in the world but 5:59 this is a game 6:00 GPU but you and I know that this is what 6:02 a GPU looks like this is one 6:06 GPU L and gentlemen dgx GPU the back of NVIDIA Networking: NVLink & Spectrum-X Ethernet 6:11 this GPU is the MV link spine and it's 6:15 right 6:17 here 6:19 this is an MV link 6:23 spine and it connects 6:26 702 gpus to each other 6:30 this is a electrical mechanical Miracle 6:34 the transceivers makes it possible for 6:37 us to drive the entire length in Copper 6:40 and as a result this switch Envy link 6:43 switch driving the envying spine in 6:46 Copper makes it possible for us to save 6:48 20 kilow in one rack 20 Kow could now be 6:53 used for processing just an incredible 6:56 achievement so this is the the MV links 7:01 spine and even this is not big 7:04 enough even this is not big enough for 7:07 AI Factory so we have to connect it all 7:09 together with very high-speed networking 7:12 well we have two types of networking we 7:14 have infiniband which has been used uh 7:16 in supercomputing and AI factories all 7:19 over the world and it is growing 7:22 incredibly fast for us however not every 7:24 data center can handle infiniband 7:27 because they've already invested their 7:28 ecosystem in ethernet for too long and 7:31 so what we've done is we've brought the 7:33 capabilities of infiniband to the 7:36 ethernet architecture which is 7:38 incredibly hard ethernet was designed 7:41 for high average 7:45 throughput because every single note 7:47 every single computer is connected to a 7:49 different person on the internet and 7:51 most of the communications is the data 7:53 center with somebody on the other side 7:55 of the internet however deep learning in 7:57 AI factories 7:59 the gpus are not communicating with 8:02 people on the internet they're 8:03 communicating with each other because 8:04 they're all they're collecting partial 8:07 products and they have to reduce it and 8:10 then redistribute it chunks of partial 8:12 products reduction 8:14 redistribution that traffic is 8:17 incredibly bursty and it is not the 8:20 average throughput that matters it's the 8:23 last arrival that matters is whoever 8:26 gives me the answer last 8:29 okay ethernet has no provision for that 8:33 and so there are several things that we 8:34 had to create we created an end to-end 8:37 architecture so that the the Nick and 8:39 the switch can communicate and we 8:42 applied four different Technologies to 8:44 make this possible number one Nvidia has 8:46 the world's most advanced RDMA and so 8:49 now we have the ability to have a 8:51 network level RDMA for ethernet that is 8:53 incredibly great number two we have 8:56 congestion control the switch does 8:59 telemet Tre at all times incredibly fast 9:01 and whenever the Knicks are sending too 9:03 much information we can tell them to 9:05 back off so that it doesn't create 9:07 hotspots number three adaptive routing 9:10 ethernet needs to transmit and receive 9:13 in 9:14 order we see 9:17 congestions or we see uh ports that are 9:21 not currently being used irrespective of 9:23 the ordering we will send it to the 9:25 available ports and Bluefield on the 9:28 other end reorders it so that it comes 9:31 back in order that adaptive routing 9:34 incredibly powerful and then lastly 9:36 noise 9:37 isolation there's more than one model 9:40 being trained or something causes the 9:42 last arrival to end up too late it 9:46 really slows down to training well 9:48 overall remember you have built A5 9:51 billion or3 billion Data Center and 9:54 you're using this for training if the 9:56 training time was 20% longer 10:00 the $5 billion data center is 10:03 effectively like a $6 billion data 10:05 center so the cost impact is quite 10:09 High ethernet with Spectrum X basically 10:13 allows us to improve the performance so 10:15 much that the network is basically free 10:18 and so this is really quite an 10:19 achievement we're very we have a whole 10:22 pipeline of ethernet products behind us 10:24 this is Spectrum 10:26 x800 it is uh 10:28 51.2 terabits per second the next one 10:31 coming is 512 Ric is one year from now 10:34 512 Ric and that's called Spectrum x800 10:37 Ultra and the one after that is x1600 10:41 but the important idea is this x800 is 10:44 designed for tens of thousands of gpus 10:48 x800 ultra is designed for hundreds of 10:51 thousands of gpus and x1600 is designed 10:55 for millions of gpus the days of 10:58 millions of GPU data centers are coming 11:02 and the reason for that is very simple 11:04 of course we want to train much larger 11:06 models but very importantly in the 11:09 future almost every interaction you have 11:11 with the Internet or with a computer 11:14 will likely have a generative AI running 11:16 in the cloud somewhere and that 11:18 generative AI is working with you 11:21 interacting with you generating videos 11:24 or images or text or maybe a digital 11:27 human and so you're interacting with 11:29 your computer almost all the time and 11:31 there's always a generative AI connected 11:34 to that some of it is on Prem some of it 11:36 is on your device and a lot of it could 11:39 be in the cloud these generative AIS 11:41 will also do a lot of reasoning 11:43 capability instead of just one shot 11:45 answers they might iterate on answers so 11:48 that it improve the quality of the 11:50 answer before they give it to you and so 11:52 the amount of generation we're going to 11:53 do in the future is going to be 11:55 extraordinary let me talk about what's NVIDIA Robotics: Self-driving Cars & Humanoids 11:57 next the next way wave of AI is physical 12:02 ai ai that understands the laws of 12:04 physics they have to of course have 12:06 excellent cognitive capabilities so they 12:08 can understand us understand what we 12:11 asked and perform the tasks of course 12:14 when I say robotics there's a humanoid 12:17 robotics that's usually the 12:18 representation of that but that's not at 12:21 all true everything is going to be 12:24 robotic all of the factories will be 12:26 robotic the factories will orchestrate 12:29 robots and those robots will be building 12:32 products that are 12:34 robotic robots interacting with robots 12:38 building products that are robotic well 12:42 in order for us to do that we need to 12:44 make some breakthroughs first we're 12:46 going to create platforms for each type 12:48 of robotic systems one for robotic 12:51 factories and warehouses one for robots 12:54 that manipulate things one for robots 12:58 that move and one for uh robots that are 13:01 humanoid and so each one of these 13:03 robotic robotics platform is like almost 13:07 everything else we do a computer 13:09 acceleration libraries and pre-train 13:11 models computers acceleration libraries 13:14 pre-train models and we test everything 13:16 we train everything and integrate 13:19 everything inside Omniverse where 13:21 Omniverse is where robots learn how to 13:23 be robots now of course the ecosystem of 13:27 robotic warehouses is really really comp 13:29 complex it takes a lot of companies a 13:32 lot of tools a lot of technology to 13:34 build a modern warehouse and warehouses 13:36 are increasingly robotic one of these 13:38 days will be fully robotic now let's 13:40 talk about factories factories has a 13:42 completely different ecosystem a robotic 13:45 Factory is designed with three computers 13:47 train the AI on Nvidia AI you have the 13:51 robot running on the PLC systems uh for 13:54 orchestrating the factories and then you 13:56 of course simulate everything inside on 13:59 Universe well the robotic arm and the 14:01 robotic amrs are also the same way three 14:04 computer systems and we provide the 14:06 computer the acceleration layers and 14:08 pre-train uh pre-train AI models we've 14:11 connected Nvidia manipulator and Nvidia 14:15 Omniverse with seens the world's leading 14:18 Industrial Automation software and 14:19 systems company this is really a 14:21 fantastic partnership and they're 14:22 working on factories all over the world 14:25 and that's the factory the robots inside 14:29 and of course all the products going to 14:30 be robotics there are two very high 14:32 volume robotics products one of course 14:35 is the self-driving car or cars that 14:38 have a great deal of autonomous 14:39 capability Nvidia again builds the 14:41 entire stack next year we're going to go 14:43 to production with the Mercedes Fleet 14:45 and after that in 2026 the jlr fleet the 14:49 next high volume robotics product that's 14:52 going to be manufactured by robotic 14:54 factories with robots inside will likely 14:57 be humanoid robots and this has great 15:01 progress in recent years in both the 15:04 cognitive capability because of 15:06 foundation models and also the world 15:09 understanding capability that we're in 15:10 the process of developing I'm really 15:12 excited about this area because 15:14 obviously the easiest robot to adapt 15:16 into the world are human robots because 15:18 we built the world for us we also have 15:21 the most amount of data to train these 15:23 robots than other types of robots 15:25 because we have the same uh physique and 15:28 so the the amount of training data we 15:30 can provide through demonstration 15:31 capabilities and video capabilities is 15:33 going to be really great and so we're 15:34 going to see a lot of progress in this 15:36 area well I think we have um some robots 15:40 that we' like to uh 15:45 welcome here we go about my 15:50 size and we have we have some friends to 15:52 join us so the fut the future of robot 15:56 robotics is here the next wave of AOS 15:59 and and of course you know Taiwan 16:03 builds computers with 16:06 keyboards you build computers for your 16:09 pocket you build computers for data 16:11 centers in the cloud in the future 16:14 you're going to build computers that 16:16 walk and as as it turns out uh the 16:19 technology is very similar to the 16:21 technology of building um all of the 16:23 other computers that you already built 16:24 today so this is going to be a really 16:26 extraordinary uh Journey for us 16:30 thank you all for coming have a great 16:31 comput text Ticker Symbol: YOU 378K subscribers Videos About Twitter 📢 Membership 🔥 Patreon 🙏 354 Comments rongmaw lin Add a comment... Pinned by Ticker Symbol: YOU @TickerSymbolYOU 2 weeks ago Nvidia's AI accelerators are insane, which is why NVDA was one of my top stocks for 2024. » See My Top Stocks for 2024: https://www.youtube.com/watch?v=jG0gBi7Ps4U » Get My AI Stocks & Price Targets: https://www.tickersymbolyou.com/survey 40 Reply 6 replies @snapadam 2 weeks ago The year is 2024 and I'm spending my Sunday evening watching Nvidia keynote, beats the hell out of Hollywood right at this moment. 124 Reply Ticker Symbol: YOU · 5 replies @hobbies222 2 weeks ago Nvida will soon be the number 1 company in the world 230 Reply 21 replies @SprakanaKerum 2 weeks ago Protect this man's company from China at all costs 6 Reply @paulokoth5126 2 weeks ago Nvidia is phenomenal. A computing revolution! 100 Reply @cosmicg3610 2 weeks ago I love how he talks about these GPUs like a proud father talks about his children. I mean, it really feels like he is not just "the boss" of this company, he is not there only for the money; Jensen is genuinely excited about both the marvels they accomplish at NVidia, and about how the world will be able to use their products as well. 29 Reply 1 reply @Fellow117 2 weeks ago Nvda to the moon! 77 Reply 2 replies @Nvidia-Lover 2 weeks ago "The days of millions of GPU of data centers are coming..." 69 Reply 3 replies @amirmansoor7578 2 weeks ago Agreed. Road map is very clear and looks like Nvidia stock will double after split. 25 Reply @Nvidia-Lover 2 weeks ago Does anyone here remember Alex's youtube video made on Jan 15, 2023 "Nvidia: The Most Powerful AI Company Ever? (NVDA Stock)". It is happening right now! 41 Reply @robertcosta9272 2 weeks ago Keep the NVDA videos coming! 30 Reply Ticker Symbol: YOU · 4 replies @user-ef1ng8th5l 2 weeks ago Thank you for this video!! Jensen Huang is def a genius visionary.. he saw this about 15 years ago & is already living in a future that prob no one alive can see. (remember when nvdia was sold off called an overpriced gaming stock) he already knew this.. In a recent interview .. when asked what keeps driving him .. his answer .."I have nothing else to do" Hilarious!! In same interview he was talking about how he is the best at setting tables!! Omg so humble .. 19 Reply @autofashi0n 2 weeks ago I got 50 shares at around 1K and I don’t mind getting in that high, I now after splitting is going to go up a lot higher. 33 Reply 2 replies @axiom66 2 weeks ago Jensen and Nvidia are the best. In this presentation everything Jensen shows is real upgradable hardware in production. 7 Reply @dennisboyd1712 2 weeks ago This Future is NOW Buy NVIDIA Stock NOW 5 Reply @2KRAW 2 weeks ago Thanks brother, You got me thru the murky waters of investing. 22 Reply @davidsoncl1 2 weeks ago NVIDIA chip upgrade annually? Three to four years used to be normal, correct? Mind blowing! 30 Reply Ticker Symbol: YOU · 4 replies @r.7442 2 weeks ago No idea what he's talking about but im hyped 18 Reply @CRBarchager 2 weeks ago Thank you for this supercut! 15 Reply @midwestcannabis 2 weeks ago I'm scared my 1330 strike price Cc will be breached this Friday after seeing this! 28 Reply 6 replies @MrSchweppes 2 weeks ago AGI and ASI in 3-5 years. Jensen is killing it! 9 Reply @Mark-tm2zu 2 weeks ago always incredible and informative to hear Jenson. 10 Reply @atticus-rex 2 weeks ago (edited) I was there. It felt like we were witnessing and on the razors edge of a tektonic paradigm shift. All of this in a world that feels like it's on the brink. Maybe this is how it feels before evolutionary jumps. Maybe this is how the apes felt at Black Monolith. 8 Reply Ticker Symbol: YOU · 3 replies @408Machine 2 weeks ago I sold a large chunk of my Nvidia position between 600-650 after a 3X because I fell for the hype that it was due for a correction and now I’m using DCA to buy back into it. Lesson I learned is don’t sell an asset while it continues to grow. I would’ve been better off waiting for a pullback before selling a portion 7 Reply 3 replies @roflc0re 2 weeks ago Nvidia will be 10t$ in 5 years 76 Reply 8 replies @robrider838 2 weeks ago This is real. This isn't like buying some speculative mining or biotech stock. 3 Reply @B.A.W. 2 weeks ago Remember those pics of hard drives from de 50/60's in the back of a truck? Imagine the processing power/size 30 years from now. I think in 50 years or less we're just gonna be telling computers "do this, do that, build this, build that". Self-built/repair machines that will help us through the universe. 11 Reply 2 replies @kuroallen6419 2 weeks ago That CEO of NVIDIA is really looking to the future XD. 5 Reply @crankybuzzard6867 2 weeks ago Terrific. Right place. RIGHT TIME!!! 4 Reply @jeffreyromanelli1708 2 weeks ago Way to go, Jensen! I wish all the success to you and your company. Reply @carmellejean-paul4334 2 weeks ago This video had me so excited!! Jensen Huang is such a good teacher! Every time he speaks he makes complex topics very easy to understand and more importantly exciting! The humanoid robots have me most excited!!! Women around the world will be most excited to finally have our Rosie from the Jetsons- humanoid robots that cooks and cleans in our homes and with AI and all the advances we are making in robotics I see that coming soon! Reply @extremeencounter7458 2 weeks ago Remember, you have until June 6th to really get in before the split. Split happens after market close on Friday, June 7th 4 Reply @tranquilitymaster139 2 weeks ago that black jacket gets better and better everyday 2 Reply @ronmatthews2164 2 weeks ago 6 months ago I sold my only stock of 6,300 Tesla shares and I bought my only stock now 4,404 Nvidia shares. I have been very lucky only owning the best stock I could find. I think the Nvidia stock price is still way too low. Everyone who last year said Meta is now too expensive when the stock went up to $ 220 is the same morons who are also now saying Nvidia stock is too expensive to buy now. Nvidia is only in their second inning of their growth. Nobody including AMD is even close and will probably never catch up to Nvidia in our lifetime. Don't overthink it buying Nvidia stock now is a simple no brainer. Selling a share now of Nvidia to buy a different stock is like trading a pound of gold for a pound of silver even up. 13 Reply 5 replies @DanielBennett-fm3jj 2 weeks ago Incredible cant wait to see the future with Nividia. 3 Reply @frankkassner1219 2 weeks ago thank you @Alex , great job, again. Love your supercuts. 5 Reply @carlmazziotti221 2 weeks ago You did an amazing job narrating this video! 5 Reply @WeirdInfoTV 2 weeks ago Thank you for editing these videos 1 Reply @GlobalMan-nr3hq 2 weeks ago I'm waiting for the humanoid robots. That will spike the next run. 10 Reply 1 reply @user-vh1td2tw1b 2 weeks ago Alex, Im so grateful for your podcasts! Many blessings to you and your loved ones. I’m buying more share bc these guys are incredible !!! No competition so much tech, great finances. Seriously I do t think there is anything like this on the market 4 Reply @rommelborja4679 2 weeks ago I’m buying more Nvidia 2 Reply @tompham8660 2 weeks ago Best CEO 4 Reply 1 reply @wyattb3138 2 weeks ago Seems oddly dystopian. Will revisit next year. 3 Reply @karimmourabitamari6540 2 weeks ago Wow. Buy this whatever the price. Soon these robots making robots interacting with robots will figure it out and buy NVDA as well! 2 Reply @dialac1 2 weeks ago Thanks for this. I started watching the full keynote but I stopped cos of the length 7 Reply @Mr.Bombastisch 2 weeks ago I am somewhat frightened of all this, sounds like a prephase of SKYNET. The technology is moving so much faster then we as society can process on what is actually happening…. 2 Reply 1 reply @Rhino5427 2 weeks ago Interesting to me how he said "taiwan builds", that statement alone upsets china 1 Reply @johnouellet4099 2 weeks ago This company is going to skyrocket, buy in now or get left behind… 2 Reply @dadehax0r 2 weeks ago Damn screw 10T projections, they're going to be the first 1Q market cap company. 4 Reply Ticker Symbol: YOU · 2 replies @Turboy65 2 weeks ago Generative AI kind of scares me because we've already seen that it will mirror its training and rules given to it. We've already seen generative AIs that have obvious, verifiable biases and we've seen them LIE. Whatever we make, we MUST ensure that humans can always kill the power to an AI gone rogue. 1 Reply @gxk 2 weeks ago NVIDIA’s Blackwell platform has officially declared Moore’s Law dead 2:51 2 Reply @TribecaSam 2 weeks ago Huang is the new Jobs 4 Reply @TribecaSam 2 weeks ago $10 Trillion value by 2030 2 Reply @ADaza1015 2 weeks ago Great video as usual. Thanks. Reply @kevindittler6524 2 weeks ago Surprised that the core / backbone is still made out of copper, I guess I could see it from a cost standpoint but limited in its bandwidth. Would’ve thought it would have been fiber-optic by now on connection points and interaction levels between the systems 2 Reply @michaelho377 2 weeks ago I will be attending Computex 2024 at Nankang-anyone attending? 1 Reply @normantang9431 2 weeks ago The tech savvy Huang puts those CEOs who keep doing M&As the spin-offs to shame. 2 Reply @economicintuition 2 weeks ago (edited) Thanks Buddy! 3 Reply @Som3D 2 weeks ago Every nvidia presentation it looks the start of an AI apocalypse movie Reply @prashants17 2 weeks ago $5.00 Thanks! 7 Reply Ticker Symbol: YOU · 2 replies @viralinvestmentsjr1526 2 weeks ago " Incredible Architecture. " Even people with the highest level of knowledge are baffled at what this truly means. 1 Reply @marga23 2 weeks ago Yes, with much respect I will like to know if Jensen, J is married! He is my boyfriend but he doesn’t know it yet 1 Reply @Lugmillord 2 weeks ago It won't take long until Jensen becomes the richest person on the planet. 4 Reply 1 reply @robertmccamman4630 2 weeks ago I only wish I could buy more Nvidia stock but I am maxed out. 1 Reply @closertothetruth9209 8 days ago will be used as a weapon in so many ways Reply @West14011 2 weeks ago When will Poet Technologies replace that huge “spine”? Hopefully sooner than later. 1 Reply @bobtarmac1828 2 weeks ago With new swell robotics everywhere, Ai jobloss is the only thing I worry about anymore. Anyone else feel the same? Should we cease Ai? Reply @fmartingorb 2 weeks ago With AI efficiency is king. You need huge data centers and the Electrical Grid is a blocker. The work AMD is doing in this direction is an important competitive advantage IMHO. While NVidia concentrates in raw power and technology, it does not solve this problem. Reply @AsaMelech 2 weeks ago What I hear: "Robots building robots." 2 Reply Ticker Symbol: YOU · 1 reply @thyroidnodule 2 weeks ago Thank you for pushing out all these phenomenal videos. I wonder if you're a thesis for SMCI has changed. The stock has taken a huge downturn Recently I have been dollar cost averaging in, but at this point i'm not sure it's worth continue. I would love to hear what you have to say about SMCI. Reply @geolykos 2 weeks ago Got in late at $1000 but I do believe this stock will outperform the Nasdaq the next few years. Reply Ticker Symbol: YOU · 1 reply @MatthewMS. 2 weeks ago I Rubin. Not talking about the sandwich 4 Reply Ticker Symbol: YOU · 2 replies @JasonG-qr5nn 2 weeks ago Thank you 6 Reply @dorout1 2 weeks ago I always believed in nvidia, but I never thought they would take the world by storm. Get hyped for not only their innovative technology, but for their 10:1 stock split this Friday -- 6/7/24! Reply @elitiano5637 2 weeks ago Thnks 1 Reply @maguilla 2 weeks ago Jensen needs to be protected 2 Reply @joncooke9515 2 weeks ago If you don’t own stock in this company you’re crazy. As long as China doesn’t mess this up. 2 Reply 1 reply @Pork-Chop-Express 2 weeks ago Is Jensen on the verge of becoming John D. Rockefeller? 1 Reply @freddyfudpucker6301 2 weeks ago Who will supply Nvidia the xhips and components used to construct the boards? TSM? 2 Reply 1 reply @eliottguerizec1788 2 weeks ago Beautiful vidéo, I have just one question : NVDA and ALAB are opponent or she could be makes colaboration ? Reply @ArjayAdamos 2 weeks ago 1400 pt before split damn 1 Reply @jamolamo999 2 weeks ago I’m concerned for Tesla he didn’t say Tesla full self driving cars, he said Mercedes 2025 and someone else 2026 2 Reply 2 replies @ahmoification 2 weeks ago The first company to hit 10 trillion market cap! Reply Ticker Symbol: YOU · 1 reply @KennyStratton 2 weeks ago You the best, man! Reply @josephshultz6726 2 weeks ago 1.21 jigawatts! Great Scott!! Reply @johal-johal 2 weeks ago Waiting for your nvidia software vedio 5 Reply Ticker Symbol: YOU · 2 replies @JohnDoe-so9oy 2 weeks ago My share atm will be over a thousand when it splits on Monday, I’m super pumped. It’ll be too late to resell them now and rebuy because my funds clears 3-5 days! I might buy more though if stocks pullsback some more. Reply 1 reply @jwdeepsky 2 weeks ago Everything changes everything Reply @NikiK57 2 weeks ago As a guy who watches all kinds of tech (and also gaming) YouTubers, my first reaction to all these practical demonstrations was - oh, you better not drop it, lol! (looking at you JayzTwoCents and Linus Tech tips) Reply @voodoochild420ai 7 days ago cool vid Reply @zackkirkpatrick6086 2 weeks ago I think they should design the creepy robots with Terminator teeth and red eyes. We know that would be amusing. 2 Reply @francopelucci6627 2 weeks ago Me no comprendio nothing but im buying the stock 1 Reply @skrowe 2 weeks ago Congo! Reply @uncrit1 2 weeks ago numbers and dimensions like in space and astronomy Reply @jun-wenshen7579 2 weeks ago Don’t stop Nvidia ! March forward continuously ! 1 Reply @amenemhurt8817 2 weeks ago Can't wait for the split on June 10! Reply 1 reply @Mark-tm2zu 2 weeks ago Jensen that is. Reply @calstongroup 2 weeks ago Jensen needs to watch Colossus: The Forbin Project 2 Reply 2 replies @Turboy65 2 weeks ago Yes, but can it run Crysis 2 at max settings? 1 Reply @johnsmithking6646 2 weeks ago Ok, ok, but can it run Crysis ? 2 Reply @timm910 2 weeks ago Why no 4k? It's hard to see Reply @__greg__ 2 weeks ago 1 Reply @g0d182 2 weeks ago cool Reply @gileneusz 2 weeks ago 10:53 rip my electricity bills, if there will be datacenter like that nearby, the demand for electricity will go to the roof, and via supply and demand laws, everyone would pay more. Reply @henzsy1404 2 weeks ago Nice , i was hoping for more on self driving besides working sigh Mercedes Reply @lifeissrecret2000 2 weeks ago Im so thrilled abt Ai development ,.I think we human beings will get erased by Ai some day! Reply @oceanic8156 2 weeks ago Imaging using that super GUP to dig Bitcoin Reply @tomburke8817 2 weeks ago X800 gri x199 2 Reply @mohammadrahimjamshidi79 2 weeks ago AI, in order to improve its performance and prevent undesirable consequences, must continuously interact with “effective rules and stable principles in the realm of existence”. X Mohammad Rahim Jamshidi Reply @kmsnow6292 2 weeks ago Terminator day is coming. 2 Reply @Gokutrump 2 weeks ago I’m telling my kids this was Moses 2 Reply 1 reply @rayhere7925 2 weeks ago Something changes everything. Yea, yea. We know the drill. Tiring... Reply @Tschoii90 2 weeks ago Praised be the Omnisiah! Reply @yutube5167 2 weeks ago do u think tomorrow 7th june, price will be like 1300usd? Reply 1 reply @danielgamingpc4090 2 weeks ago 1 Reply @WilliamAshleyOnline 2 weeks ago wow. Reply @k1m625 2 weeks ago technology used to get smaller and smaller.....now its getting bigger and bigger Reply @PeeP_Gainz 2 weeks ago My comments are historical facts, just need timing to become accurate. Nvidia: Top Company on the planet Q4 2024! Reply @guyguy12385 17 hours ago not having tesla on the list is hilarious will be fun to look back at this Reply Ticker Symbol: YOU · 2 replies @lauriebarns9901 2 weeks ago How do we deal with the transfer of wealth creation from humans to AI resources. Without a level of planning that we have never seen before; it going to make the Luddites look like real amateurs. As the architects of this structural shift, the AI companies really need to give this serious thought. Reply @bigdhav 2 weeks ago Cerebras chip is the largest chip in production... Reply @juancedano5839 2 weeks ago If we gave Jensen a place on Star Trek the TV show, he would be who. I say Q. 1 Reply @eamonnmcdonagh2743 2 weeks ago He has got a definite Jimmy Savile vibe. Reply @Josephkerr101 2 weeks ago I just envisioned Linus giving this speech. He would have dropped literally everything and it would have been hilarious. Reply @jth5542 2 weeks ago This is what should have happened to Intel if they didn't get complacent with innovation. Reply @RubiconV 2 weeks ago The stock YOU has gone down from 34 to 17. Just saying. 1 Reply @DeepZenMind 2 weeks ago Jensen keeps accelerating the compute technology; the accelerated compute keeps accelerating the innovation; the accelerated innovation keeps accelerating NVIDIA's own growth. As a result, NVIDIA's stock price keeps rising and accelerating. 3 Reply @LeicaM11 2 weeks ago So 20000 Teraflops at FP4 are 5000 Teraflops at FP16, to compare it in reality or 5 Petaflops. Reply @eric78730 2 weeks ago when was this presentation? Thanks! Reply Ticker Symbol: YOU · 1 reply @michaelho377 2 weeks ago Nvidia is Awesome Reply @earlwilliams5473 2 weeks ago One day all that will fit on a microchip. Reply @sethcockram 2 weeks ago will nvda go back up beacuse stock is down in 224 I hope u guys can come back 1 Reply Ticker Symbol: YOU · 3 replies @AJ-py2kt 2 weeks ago Reply @6lack5ushi 2 weeks ago Woooow 1 Reply @SovrinWealth 2 weeks ago Amazing - "It's whoever gives me their answer last" - always "out of the box" thinking attitude which allows "quantum jumping" for all of humanity despite most of us being procrastinators and disbelievers when the obvious is already tickling our noses... Reply @user-gk7iq6if9q 2 weeks ago WOW 1 Reply @Brismo7 2 weeks ago 2:40 why they got to make the AI motherboard look like a freakin skull...??? Reply @pascalsierens6299 2 weeks ago Reply @TrasThienTien 2 weeks ago The stock has skyrocketed Reply Ticker Symbol: YOU · 1 reply @Woter_X 2 weeks ago So would this be a good thing for Minecraft? Reply @keerthikumarc5894 2 weeks ago Hi bro can i buy now before split. 1 Reply 1 reply @ossurskarp8517 2 weeks ago I completed your survey but didn´t receive what I felt was promised... (("Get My AI Stocks & Price Targets) Reply @cinsationalcinema1776 2 weeks ago I wonder what’s going on in the top secret section of their company? I can only imagine something like terminator is being created... just saying. Allegedly of course Reply @TriniNY 2 weeks ago (edited) 13:03 Terminator units…….kidding Reply @SilviaD-zg5ov 2 weeks ago NVDA my favorite stock the CEO is a genious NVDA going to the moon very soon Reply @Emc2Eggs 2 weeks ago Let's hear second niece Lisa Su mañana 1 Reply @DK-ox7ze 2 weeks ago Are you buying more Nvidia right now? Or would you wait because of the recent peak in its price? Reply @stevenanderson9660 2 weeks ago I’m going to be rich 1 Reply @LetsG0Champ 2 weeks ago RIP Engineers and millions of other jobs. 1 Reply 2 replies @Capeau 2 weeks ago This guy still pumping? I saw not much new... but sure blackwell performance looks nice. Reply @tagsoneveryting 2 weeks ago And......Terminator. Reply @gustavocoelho11 2 weeks ago Congrats for your picks on U and PATH, mate! Reply Ticker Symbol: YOU · 4 replies @au81512 2 weeks ago Yes, Chinese characters. Nice. 1 Reply @jimgrimm2874 2 weeks ago I'll be able to run GTA 6 @24 fps! Reply @MTXSHO9732vV8SHO 2 weeks ago 8:05 Herein lies why POET Technologies hardware is so important... You (probably) read it here, first. Reply @williamreed7099 2 weeks ago SCARY! I mean, snake skin jacket and the all black with the motherboard configured like a skull or some kinda creature. I mean, hey, I guess you have to over sell it! Reply @AjitMD 2 weeks ago The addressable Global Economy is huge… and growing fast. Not just USA, but also rest of Americas, Europe, Asia, Middle East. One significant restriction is China and its economy is huge. Yet, some sales of slower versions are taking place. I think that commercial companies like Ali Baba, Bidu will be able to use data centers under USA control. Recurring revenues in AI software, services are important to sustained valuation that could exceed MSFT? Reply @aprillynmoon9225 2 weeks ago Hahaha hoping for a rally on Monday after seeing this video Reply @OverLordGoldDragon 2 weeks ago Do you intend to address Groq, and its threat to Nvidia? Reply @pauldannelachica2388 2 weeks ago Reply @pje11111 2 weeks ago how about you not staing silent anymore Reply @rommelborja4679 2 weeks ago I got all my retirement money on Nvidia Reply @andrewereynolds5840 2 weeks ago Trading is a global poker & chess game that is executed thru a auction system (order flow thru bid/ask). Dynamic profession. Reply @Travesten 2 weeks ago No Cereno Scientific will CHANGE everything 1 Reply @newjustice1 2 weeks ago I think nvidia has agi add its making there chips Reply @TimAZ-ih7yb 2 weeks ago All this “magic”, and they’re still not able to demonstrate anything useful or profound in the world of human activities. NVDA apparently excels at surfing this wave of AI hype, good for them. Reply @cynthialopez2201 2 weeks ago so robots in warehouses run by robots and robots building robots okay got it! Reply @baltimorehoodboss1157 2 weeks ago They cannot sell to china...but there are Chinese translations on his presentation. Rubbing it in? Reply @MassimoPardini 2 weeks ago How high can go this stock go ? Write what you believe below ️ Reply @minefacex 2 weeks ago Blackwell is literally not the largest chip that the world makes Jensen XD Reply Ticker Symbol: YOU · 2 replies @nithinbhandari3075 2 weeks ago 3:09 I am not able to understand the comparison, How nvidia can compare FP16, with FP8 and FP4. Like hopper has FP8 with 4000 TOPS, and Backwell has 20000 TOPS with FP4. So how can Nvidia compare , FP16 with FP8 and FP4. If we check with FP4 * 4 = FP16, then FP4 = 20000 TOPS is equal to FP16 = 5000 TOPS. Or they can even say, FP1 80000 TOPS. (I have limited understanding of topic related to GPU). 4 Reply 3 replies @Metaplayer 2 weeks ago Great video, but can we not hold back on these empty "this changes everything" clickbaity titles? These words have lost all meaning in the context of language models. It makes you look less serious. Common, you are better than that 1 Reply Ticker Symbol: YOU · 1 reply @Lightsydephil 2 weeks ago Nvidia killer AMD's keynote begins tomorrow Reply @aaap3875 2 weeks ago Nvidia owner = Taiwanese American....Morris Chang TSMC= Taiwanese American AMD Owner = Taiwanese American. Foxconn= Taiwanese. All PC Brands are Taiwanese made(Acer, Asus, Apple, Lenovo, Dell, HP, MSI) Reply @J-Kimble 2 weeks ago Nvidia is a bubble. Last as long as other NPUs get scaleable, which is around 1-2 years. I would stay away unless you really know the accelerator market. Reply @niklas7355 2 weeks ago Nvidia bubble will bust!!! Go out now Reply @taiwantaipei3913 2 weeks ago How he keeps mentioning a Taiwanese woman who sells fruits in night market for 43 years he asks People here to buy some from her Reply @xwolf9913 2 weeks ago Sky net lol Reply @mrhassell 8 days ago Changes nothing. Next GPU will be AMD. Reply @SkywalkerPaul 2 weeks ago Say hello to massive unemployment Reply @nguyenlong-rh3uc 2 weeks ago (edited) The world goes smaller lighter, he goes larger bigger how bigger his machine can compair with historistic computer in musium what a move nvidia Reply @baltimorehoodboss1157 2 weeks ago Robots interacting with Robots in factories...no need for Chinese or Mexican cheap labor Reply @minhucle9593 2 days ago Very non Reply @GerryPrompt 2 weeks ago Way to just re-upload a video... Reply Ticker Symbol: YOU · 1 reply @jasecee8992 2 weeks ago Change the fucking DESKTOP PANEL first! It is the step in the right direction. Pathetic we still have this disastrous piece of software. Bland, slow, laggy, lame. everything bad. Reply @firstsonofthesea7938 2 weeks ago The greater the performance we drive up, The greater the cost decline. Bullshit......Bullshit Reply @ZsZw 2 weeks ago $NVDA Tomorrow let’s Reply @m89hu 2 weeks ago First? 2 Reply Ticker Symbol: YOU · 1 reply @stevenjordan5140 2 weeks ago You spoke too soon. AMD just dropped a A bomb on AI. It coming 1-2 Q latter is 150% better and fits into existing sockets dramatically lowering refresh costs and TCO. AMD take the crown 6 months after Blackwell at 25% of the refresh costs and 50% of the new system install costs, all while being cheaper to port and program using open source software and industry standard interconnect. Reply @BigErnieBrown 2 weeks ago My friend Harry has a brother Larry. In five days from now he is gonna marry. Reply @jfry3276 2 weeks ago He’s just a salesman keeping ya’ll drooling over “ooga booga”, “allegedly”, big word-technicals you don’t even understand… I’m not sure he does. Sounds like he makes half of this up. I mean, how much can one guy talk about chips… he has to make stuff up, at some point. Reply @pawnslinger1 2 weeks ago Jensen, Jensen, Jensen, where are the humans in your plans? When all the jobs are done by AI robots, where do humans fit into this? Jensen, have you considered the world you are enabling? How will the vast majority of people earn a living? Even the artistic world is being invaded by AI. Will any pursuit be left for humans? You are so optimistic and enthusiastic and you clearly believe in what you are creating. But have you thought about us?? The humans. Or have you left that for someone else to worry about. It is all above your pay grade, huh? Because it can be done, maybe someone else would do it, you rush in to this ultra modern world you are creating. You and Oppenheimer have a lot in common. It finally dawned on him that he had become the destroyer.... will that be how you are remembered, Jensen? No, maybe I am too severe... you are the new James Watt. James was responsible for much of what we see in our world, the good and the bad. Consider what his creations have done to the world. The profound changes.... and we have yet to see all the effects that James Watt will have on the world.... just consider the tornado season seems year round now. Will your creation create a whirlwind too? A whirlwind that will sweep all before it? Just think - Jensen Huang... the next James Watt. Your name will live on... Reply 1 reply @darkashes9953 2 weeks ago (edited) Blackwell is still way to week and way too slow and un efficient. Reply 5 replies @assassinipo3082 2 weeks ago Reply @J-Kimble 2 weeks ago Nvidia is a bubble. Last as long as other NPUs get scaleable, which is around 1-2 years. I would stay away unless you really know the accelerator market. Reply 1 reply