Nvidia CEO Huang: Our country should invest in AI and apply regulations where it makes sense

you get the chance to sit down with one of the most important business people on Earth Jensen Wong the founder president and CEO of Nvidia you take as much time as they'll give you even though the stock has come down from His Highest it's still pivotal right now so we had lots to talk about take a look now we are trying to figure out whether AI will always be derivative when I put in a question in the end just it's it it it just just it's scrapes everything but it scrapes everything I don't want I

want new I want something original and not derivative will they able to put it all together and actually suggest that this would be a better way to look at it turns out that all of that Foundation knowledge is just Foundation knowledge what we really want the AI to do is to learn how to solve problems learn how to reason apply rules apply knowledge break a problem down one by one by one step by step by step and solve problems that's intelligence but you can't teach an AI that skill of solving problems you can't teach AI

reasoning without first having a foundation knowledge of what are numbers how do words work together what's vocabulary what's syntax what's grammar so that it could learn some basic knowledge once it has the basic knowledge like we have ai today notice how quickly we're teaching them how to reason right and so it's learning how to reason it's learning how to solve problems it's learning how to solve math problems geometry problems prove theorems prove logic it's it's starting to learn those things by itself and so it's really fantastic well this next level even now when you look

at uh a deeper search from gbt when you look at Gro they have opinions yeah uh they're thoughtful they're not yes men anymore well they they'll argue with you right yeah that's happening now though yeah yeah it's fantastic and how that you did you know what would be happen next year last year Well I you know I'm in the industry so so we see progress and I can tell you I can tell you that that AI is going to progress very quickly over the next several years and and uh the answers that it provides it

will be uh surely more accurate surely uh much more contextually relevant and uh it'll be able to answer and solve problems that that um quite useful to us what can our nation and our government do to further what you're talking about um for well I think that that our country should also invest in AI uh you know we should of course be um uh talk about talk about the importance of the technology uh apply regulation where it makes sense um but also uh invest in using AI across the across the country in every every why

don't we have a strategic AI Reserve we have one for something as meretricious as Bitcoin we should at least build the AI infrastructure so all of our scientists in United States can use uh the AI infrastructure and um uh however it comes about whether it's a strategic Reserve um but the country needs to invest in an AI infrastructure now when we talk about this we almost immediately talk about the Chinese and almost a Sputnik like race in 1957 are we being too uh too small mind and we shouldn't be thinking about the amount of business

you do with China maybe the things are are are are much further along than that mentally are we constraining ourselves with these discussions about politics and tariffs are these too small if uh I think it's safe to say that we can't hold any country back or anyone back uh in advancing and developing intelligence and surely AI is just digital intelligence okay and so I think everyone is going to develop it uh if everyone's going to be developing it let's make sure that they develop it on American Technology American architecture and the American Standards and so

so that's what I would Advocate um and then secondarily I would say let's make sure that we are uh creating the conditions and investing in a way that allows our country to be far far far ahead and and so uh everything that we can do to speed up our advancement everything we could do to speed up education here um uh startups here uh Economic Development here using artificial intelligence uh we absolutely should no but when there was this deep seek issue your stock took a $600 billion hit I tried to reach you I was told

it was quiet period that we could have easily explained that this is actually good forid why not have said that during that period uh we were during a quiet period however you still came out two days before it was 25 days and then you did it in a video two days before well next time next time you should contact me directly and and uh I'll come out and explain it because it's a good thing yeah R1 deep seek was a fantastic thing and the reason for that is because R1 is the first open- sourced reasoning

model what makes R1 incredible is that it reasons that's why the answer is so good and it breaks the problem down step by step it uh asks itself while it's thinking it comes up with several different options for the answer it might actually verify that the answer is correct it solves a qu quadratic equation takes the number plugs it back in and makes sure that it actually solves the the equation and so so it's a reasoning AI this reasoning AI consumes a 100 times more compute than a non- reasoning a and that's you it was

exactly the opposite it was the exact opposite conclusion that everybody had it actually consumes a 100 times more Computing well it it's it's kind of amazing because that the people are threatened by Quantum Quantum may use a thousand times more Compu well we're going to have to go help all the quantum computer companies and this is going to be the first conference at GTC where we have a Quantum day and so all of my friends in the quantum industry is going to are going to come and we're working together to create classical Quantum hybrid computers

because a computer can't just be Quantum and we like Quantum to to uh uh be able to accelerate many algorithms and so create creating a classical uh quantum computer using this architecture we call cudak Q uh it's been adopted by the two years 5 years 10 years well you know I'm not I'm not making any more predictions no right and pred yeah but I but I do know that that all of our colleagues are working as fast as they can I know they're making excellent progress and we're working with everybody to to uh accelerate the

developments of quantum but right now we have we $150 billion dollar of AI infrastructure into trillions of dollars we have to go build and so we have Partnerships with black rock and Microsoft and and uh uh we're going to go get the supply chain uh and the infrastructure uh ready uh Upstream as well as Downstream and the demand is really incredible Black Rock reached out to me this morning to say look this is incredibly important Black Rock they were talking about a one gig data data center 4 40 to 50 billion Disney reached out to

me today Cisco talk to GM uh crowd strike these Partnerships that are ey glazed over Partnerships when you have the 18 that were announced yesterday that I counted are all incredibly significant for big companies that trade publicly that want me to know how important what you're doing is with them but no one seems to understand that there could be 20 30 things that you're doing we announced several things yesterday and and let me let me let me make it very simple this way we're building AI infrastructure for the cloud that everybody knows right that's a

true that's true everybody knows that there's just delighted that all of a sudden um AI reasoning is going to require a 100 times more Computing and the demand is incredible so that that's AI Club um what we announced yesterday is that they we're going to take AI infrastructure into two new Industries one is Enterprise it half of the world's capex is an Enterprise Enterprises you know how the econom not consumer it's Enterprise and it's the IT department of Enterprise and so we have to help them integrate AI into the world's companies right and so our

partnership with Dell and HP and Accenture and service now and crowd strike and you know all these great companies ddn and net app and vast and all these great companies that we're working with are going to reinvent the Enterprise it with new AI infrastructure and then the third growth opportunity for us now is AI for robotics before you have robots you have to do AI for robotics you they have to be smart and so that's going to take mountains of infrastructure as well and so we now have to build AI infrastructure for three Industries for

the cloud right for Enterprise it and for robotics well that's in the end that's the world this is yeah we're going to we're going to be the foundation for the world the world hopefully yeah one company oh we have a we have tons of Partners look at us you know just about every company in the world's the just about every company in every industry is here it's incredible we have car companies healthc care companies of course the entire computer industry is here well I mean let's say I have that robot that's in my house it's

joining me Bing me a good cup of joe how many black well no Ruben I don't know chips that will how many will it have of you of yours well it only needs one of these oh thank heav 140,000 chip there no no it's a smaller version it has one version that sits in the computer inside the robot but don't forget there's a data center a whole AI infrastructure supercomputers teaching the AI to run on this computer to do its do its job do its tasks and so the AI That's in the data center is

gigantic and you see these systems behind us that all these systems is what it's going to take to develop the software that runs on a chip like this inside the robot and so the robot is actually the end that's the end the beginning is all of this well okay I'm I'm thinking about a world where it it feel just between this year and Le and last year the future is coming to the present they are converging right now right here fast and super fast super fast and super fast look a year ago we were talking

about generative AI where AI was able to produce text summarize documents Dynamite create images all that stuff that was fantastic last year was all about generative AI all of a sudden here we are one year later everything is about reasoning solving problems AI is doing math solving geometry you know and so they're reasoning and this reasoning capability as it applies to robots is you give a robot a job a task and it's got a reason okay what do I pick up first right what do I man maneuver it around what do I do next maybe

open a drawer and how do I put this in should I stand it up or lay it on the side it's got to reason about that can it make the bed well sure it's going to reason you said make the bed and so it's going to say okay make the bed what does that mean break it down step by step by step my exclusive with Jin Wong continues after this don't miss a second of Mad Money follow at Jim Kramer on X have a question tweet Kramer # mad mentions send Jim an email to madmoney

cnbc.com or give us a call at 18007 743 CNBC miss something head to madm money. cnbc.com