achieve model improvements with less in terms of gpus? it's a very important distinction to make that deep. seek has built a model that does inference cheaper right now at current performance levels. i think that's the important takeaway, but it hasn't proven that they can build a model that exceeds the performance of the cutting edge models that we have now in the united. states that. have been built with nvidia chips. so that might not feel like good consolation today, but i think that is the narrative over the next few weeks. what are these companies say about how they're continuing to invest and deploy large infrastructure to build these cutting edge models that will be better than deep tech over time, i believe. >> i mean, as we've already discussed on this program, the narrative around open source versus not seems to be where the action is and where this story is ultimately potentially going. you look at meta today, for example, which is up 1%, the only of the hyperscalers. i don't really include apple. it's just a different kind of

i said there are no gpus without cpus. >> i have never seen a gpu without a cpu in a system. >> nvidiaemselves. the cpu is fundamental to just about every digital device you can imagine, whether your automobile, your ring camera, your iphone, android phone. you have to have cpus. they run a lot of workloads. we are happy to combine with any accelerator. >> we listen to nvidia. okay, i have to be in his kinds of chip. his chip doesn't ship without your kind of chip. >> absolutely. there was cool announcements they made. one of them was digits as he called it, which is an ai minicomputer. a blackwell gpu where you can run ai on your desk-top. a good example of what you can get in terms of the arm cpu along with a gpu from nvidia. >> i did not know that. i think that when i aid there are a lot of people who think -- i have some knowledge. but people see -- i don't think you can outrun the cellphone industry. it doesn't need to outrun the cellphone industry. that's the old arm. >> we have expanded our business so much over the last decade, whether the data center, automobiles. security dev

that does include the gpus, looking at blackwell. around $30 million a megawatt. so $500,000, you're going to need anywhere from 12 to 14 gigawatts and that's the crux of the issue. you can spend all this money on gpus, but if you look at the data structure in the u.s., it's only 21 gigawatts. that's over 21 years. the power density on those racks is maybe less than 20 kilowatts. and that's the crux of the issue is blackwell's need 60 kilowatts if not more. so you can't use legacy data center so you need 12, 14 gigawatts of new power. >> i'm curious if you can rattle off publicly traded names, whether they're the minors, moving today or being direct beneficiaries. >> if you look at this project, they want power at scale. that's what the minors have done very well is neck to the grid. you know, one is riot. riot has a gig watt site in corsica. it's about an hour from dallas. star lord took a state wanting to push into ai. they said we're going to stop our bitcoin mining development and evaluate this opportunity. that's one we really like off the back of it because

the company was built on gpu's. he's going back to those roots. like micron searching today. its own memory chips are going to be within these gaming gpu's. to force --geoforce. it will predict what image is about to be generated and facilitate far better experiences. this is jensen huang trying to articulate yes, there are ai data centers but they are about more. there about gaming, robotocs -- robotics and autonomous vehicles. scarlet: the fact that these companies are in nvidia's orbit are seeing the orbits of being associated. what about the linkage to the white house and washington? i like how ed brought that in. have you paid homage to the incoming president? caroline: have you kissed the ring? he's willing to go. what was interesting was trump was on at mar-a-lago discussing a new $20 billion investment from the middle east to do what? invest in data centers in the u.s. those are going to have nvidia chips. he's going to be a winner. the cfo has been speaking about the jp morgan event, particularly that blackwell and ai accelerators are her

anxiety for investors questioning the thesis that the spend, the capex spend on companies for nvidia gpusll hold up as a thesis around the support for this ai revolution. european stocks taking a hammering in terms of futures at least, looking to lose a full percentage point. ftse 100 futures in the u.k. pointing to losses of 0.5%. s&p futures looking to lose around 79 points. nasdaq futures with that tech heavy focus, currently pointing to losses eye-watering of around 500 points, lower 2.3% on a big week for tech earnings with the big members of the mag seven, five of the seven reporting this week including meta, who caused concerns friday. talking about that capex spend themselves. let's focus on dollar strength. yields coming down across the treasury curve leading up to the fed meeting wednesday. 4.58 on the 10-year. euro-dollar at 1.04. the single currency down 0.2%. brent taking a knock linked to the terror threats against colombia, down 0.8%, and bitcoin back below $100,000, there you go profit taking the cryptocurrency down four point 5% in today's session. let's get to the deepse

those capabilities are built-in to invidious gpu inward deep-sea cassette itself apart is by lowering cost and that is why the market is reacting the way it is. it took less than 6 million according to deep-sea to train 700 billion perimeter model this is significantly cheaper than what the united states is spending to trade their large models the market is worried and i would argue that is the wrong readthrough this will cause the united states and big tech companies to spend more what about the open source, a lot of people saying that is a game changer as well. >> open source goes hand-in-hand with the idea the whole point was to lower cost for a.i. training in a.i. inference which is we need startups to participate. we need enterprise companies to participate we don't want the technology to be locked behind the interest of big tech, the nod to open source into driving down cost. >> if elon messes somebody is using a lot more gpu suddenly claim in the 10000 maybe 50000 also canter for nvidia sounds like you agree with this. >> i bought nvidia this morning china may be number one for

the gpu isjust a beast. yep, it's ai.hip, also known as a graphics processing unit or gpu. they'll still be found at data centres, but they're now coming to a laptop near you, delivering these movie—like images for gamers, developers and content creators, huang said. many in the audience were impressed. previously, when we looked at all these kind of graphics, we think of them as pre—rendered, as some, like, disney—grade or, like, dreamworks—grade animation studio, but now they can do this in real time. in addition to promoting nvidia's latest products, jensen huang also seemed keen to promote ai in general at a time when many consumers still don't see the use case for it in their own lives. it used to be ask a question, answers start spewing out. in the future, you ask a question, a whole bunch of models are going to be working in the background. huang is describing agentic ai, a term used to describe the automation of everyday tasks. the plan is for nvidia's new chip to power its rise. people are going to have agents th

we got used to the idea that they are this stalwart provider of gpu's and data centers and what they announced and what these images are showing our them taking that work and making high-performance gp use for the consumer and for developers that are a fraction of the price of the prior generation. data centers are like 100 billion for the full year but in a future where ai is everywhere i think he has his eye on other markets. >> it's fascinating as a diversifier. does that speak to concern for you amongst nvidia that maybe it is overexposed and this is a key test for analysts in terms of what one has to say about the push out into other segments >> it will be interesting to see the wall street mood tomorrow, whether they got any of it. it was not much focused on this is the brain the strange the ir model but everything else. what jensen means is the next phases in the physical world. nvidia has a hold in the corner of that market which is that the data centers are being used to train the underlying models that support that technology but there are many more steps to calm to build r

and do we know whether these people have 500,000 gpus or 5000 gpus? why should we trust the stories we're getting? >> well, that's a good question. i think a lot of people here are also wondering about kind of the details of this company, because it's come onto the scene quite quickly. people are wondering where the money is from. there have been some questions about whether or not some of it was government. then, you know, if you look at the research, it looks as though it's a very private company owned by the founder himself, as well as two other friends. so there are a lot of different questions about exactly how this company has been able to do this. but, you know, still, we're just gonna have to wait and see. >> you appreciate it. on another day, we might be talking about the manufacturing pmi there, which fell to a five month low. but this is the story of the day for sure. eunice yoon in beijing. take a look at the premarket. everything we've just been discussing is getting reflected in the futures which are off the session lows, turning into a

but i'm not actually convinced at the end of the day, it's two gpu vendors. i cover broadcom, who does asics, and i do think the ultimate is probably nvidia gpus and then for captive workloads at hyperscalers, it will likely be asics. they're all working on their own. and i expect asics will outgrow, in the sense they're coming from a much smaller base. you can do the math today, it's a little bit of monkey math, but i bet asics are probably low double digits. could that number be 20% of a much bigger pie in five years? yeah, i think it could. but my guess is, that's the -- the likely competition. now, yes, does it worry me? i don't think we're on the saturated part of the s-curve, right? things are still growing. so, i think the question, is it a big market or is it not? if it's a big opportunity, i think there's room for everybody. if it's not, then everybody's in trouble, right? so, i think that's the question that you want to have confidence one way or the other. i'll worry about competition later. i don't think it's the time to worry about it now. >> sta

that's the nvidia gpus that everyone wants. nvidia 800 are the ones that the chinese get because of the export ban. and what deep sea says it was able to do is create a model that outperforms some of our best models built here in america with those dumb down chips. i mean, that is a point of debate. some people say that the hedge fund of which deep tech was born out of had been sort of hoarding the better chips. but at the end of the day, kelly, it doesn't really matter. we've moved on past that. we know now what deep sea proved is that these models and open source. nonetheless, i want to stress how important it is that deep six model is. open source can be built for a fraction of the price, whether that's 6 million or 60 million. that is far less than the hundreds of millions or billions of dollars that openai, gemini, google and others have poured into this. so i think that's also maybe why you're seeing meta in the positive today, because it's also working on open source matters. in fact, i can't remember who it was. one of

was the cost to train the model costs less, and similarly, they're saying if they had access to more gpustter than it is. for the industry as we see it. yeah. that actually the end game, because it uses less energy in terms of where ai is going. yeah. the term inference to hundreds of millions it opens the market. it increases the tam. cost of compute down. there's obviously going to keep looking at those efficiencies. the new i secretary wants start 2.5% on american imports according all american imports according to reports the financial what you think of this idea? 2.5% tariffs on all the 77 77 77 the us 77 tv tv theus 77 off 77 intoihe us regardless of of i222? us regardless of of i bathe , ,4 , - of ~' bathe : :: , j: of ~' country of origin? i think it is important _ country of origin? i think it is important to _ country of origin? i think it is important to start - country of origin? i think it. is important to start i saying is important to start by saying there is no consensus and trump after the comments leaked went after the commentsleaked went on a limb to see out on a limb t

those of chip names are pushing higher because when an investment is made into a high-performance gpu a goes in combination with emory chips, other cpus, and telematics from those providers. that is the key question for jensen. i see a bit of a concern about inflation in the market this morning. jensen has a lot of worldviews, whether he will talk to me about inflation we will have to wait and see. caroline: we cannot wait for the conversation. you will see you a little bit later. let's get an investors take on the here and now, lack of his ability on blackwell or the future-proofing of the business model. cbiz investment ceo anna rathbun. it is 9:30 p.m. new york time when we heard jen-hsun huang on stage. are you in line with some of these analysts saying, not enough on the black platform? anna: yeah. there were not that many details that investors were hoping for. but if you have to talk for 90-minutes-plus i'm not sure you can fill that timeline with a lot of details about the business. plus, he is kicking off a conference. you sort of want to be a visionary and there is this litt

the bitcoin miners they have the gpus and they have the electrical power contracts. one being one of the key ones right there to transform their business to take unused capacity in particular, and make it into hyperscaler capacity. that company is under pressure from significant share holders. starboard is there and i'm aware of others as well. apply digital deal, small market caps but an interesting theme as we know how important it is to have the gpus and have the power. if bitcoin stops going up, they will probably not make much money. i'm done, jim, go ahead. >> that was great. the power shortage in the country is extraordinary. i keep talking about iceland being totally renighable and yet they are not a friend of the united states. let's bring in eli lilly ceo david ricks at the healthcare conference. you're the man of the hour. >> good morning. >> let me express condolences about the loss of your father. >> thank you so much. >> we are stunned as usual about the fact that the stock is down. and i don't want to be too pedestrian to talk about the stock because i

they focus on gpu, this could be brand new.g in that world, to compete directly with amd, to compete directly with intel, as well. so lots to go over, but those are the four main points for nvidia. >> kristina, thanks very much. we appreciate it. it's going to be a busy week on the tech front. >> yes, because as we mentioned, the tech world is focused on the consumer electronic show known as ces, but there's another big event looming, which also has caught the attention of many tech ceos. and that is the inauguration of one president, donald j. trump. >> never heard of him. >> who? >> steve kobach joining us now with more on big tech. is this fair to say, steve, a cozying up relationship? >> cozying up, courting, some would said dancing. but the latest for those courting trump ahead of this inauguration, it's microsoft and apple's ceo, tim cook. each offering different things. and it may not be the end of what they have to offer the incoming president. on friday, axios reporting that cook is donating $1 million personally to tr

if those models are so accessible, a lot more people will use them, which will drive the gpu demand theyave for training. scarlet: thank you so much. the head of protocol over at autonomys. that does it for this hour of bloomberg markets. i'm scarlet fu. balance of power is up next with kaylee and you -- kailey and joe. >> from the world of politics to the world of business, this is "balance of power." live from washington, d.c. joe: it is time to meet the press. welcome to the faster show in politics as the trump white house prepares for its first breast briefing with a newly minted secretary at the podium. i'm joe mathieu in washington. secretary caroline levitt in the west wing in a moment, the youngest ever white house press secretary hours after donald trump called for a pausing federal grants, called again for bigger tariffs. we will bring you into the briefing room and carry this life for you. the first conversation between the secretary and reporters including those representing bloomberg. a bit of a comeback in the cards. charlie pellett has the markets for us right now. charlie

systems from # gpu's to 72 gpu's these other players cannot keep up whether it's custom silicon fromadcom or amd. however, keep in mind, charles as this continues, the pricing power that nvidia has almost pricing power will com moment will occur so give them that time. i would say 2026, 2027 i would not give up on amd, but with that said, right here right now, this is nvidia's market and it continues to be. charles: i'm so happy we were able to get you on today, beth. learned a lot just from our short interview. because to your point this is the place to be, this is where the money is made and where society is changing and i'm glad we have you helping us understand it. thanks a lot. >> thanks charles. charles: you know, folks. the problem with the headlines like the one that started us off today from cnn on trump's tariff plan is they aren't breaking news. they aren't breaking news, but get ready because it's going to happen a lot. they are going to move markets and they may actually lose you money. i'll explain and try to caution everyone right after this. morikawa on 18. he is real

the transition from hopper, which was the current gpu iteration to blackwell is not as smooth as what a lot of analysts were anticipating. i say that because the ceo and cfo both said they're not going to up the guidance. they said blackwell is shipping beyond their expectations. but if they're not upping the guidance and one product is doing better than expected, the other product is not. so we have to think that going forward in a lot of every single year we have this moment of transition from one product to the next, especially nvidia creates this annual cadence. that's something important for investors to keep in mind, that customers are going to hold back and wait for the next iteration of the gpu. that's number one. number two, overall, we got a lot of robotics and gaming, but that's not near term catalyst. so that's probably why you saw more of a negative react ion right now. >> kristina, thanks. kristina partsinevelos, appreciate you on that. let's bring in malcolm etheridge of capital area planning coat. you hold apple, what do you make of these reports and the question that

he goes, you know, there are gpus and there are cpus, and when there are ever gpus, then you find arm. so i would not call it regulatory, i would not call it new. but we have a market that has animal spirits, david. and you can say the same thing on thursday as you do today and people take it up. >> you can carl, in part because the numbers being bandied about are just so enormous. now 100 billion is obviously a big number. who knows, if they really get to the 500 billion, it's possible you got the saudis now talking about 600 billion investment over four years. by the way, the saudis are always investing here. a lot of their money ends up here. so it's never clear when you make these kinds of announcements how much of it is actually new. or public capital private. we don't. >> we're still waiting for confirmation on the state news agency report. >> right. and the conversation between right. the saudi leader and. >> they're putting out some pretty big numbers. >> they're putting up some big, big numbers everywhere. >> big numbers. >> big. big microsoft by the way, 80 billion in capex

it took around 72 hours to create this example from scratch, using extremely powerful gpu.

die net have ettess te the and best they lie net heye eeeeee te the and best �*these ; nvidia models, gpuriven to have been driven to innovate. the thing is 7 they made thij they made thisvmodel on $6 versus a company like who is estimated to have $6 billion investment. e $5! hittieh ihveetreeet.—ttie— —— — interesting coming else interesting. cemihe cm. the w w. e e. ,, ,, of else ihtereetihe cert-tine cm. the e- w. e e. ,, ,, of trump heels of president trump announcing stargate which would a $5006 7 ' 535661160?" ,,, be a $500 billion infrastructure ' us. us. , ., the us. in terms of what now, there is massive shake—up going on as m but le shake—up going on as m but le sha you p going on as m but le sha you expect on as bc, but would you expect someone like totally the ff�*fi�*ef�*fie ye; gg'fii'étf�* �*fi�*§“§ " ' the way they eeh�*eehe eiee ee'e'e�*e �*e�*eeee " ' the way they are eeh�*eehe eeee ee'e'eee ee�*eeee " ' the way they are going change the way they are going it, be forced a? is hard to tell. the it is hard to tell. the popularity of the building and developing ai will call mor

. >> there's one gpu. >> that's fueling. >> ai and it's nvidia. and when you think about these build outs, when we talk about the enterprise, the 2 trillion. >> of. >> ai capex in the next three years, that doesn't change. >> from. >> deep sea. we're talking about 1/1000 of ultimately the build out that you're going to need for a lot of these enterprise use cases. you look. >> at names like. >> palantir, you look at the hyperscalers. >> i could argue with inferencing. if you actually. >> think costs are coming down, it's. >> actually bullish. for even some. >> of the software players names like salesforce, servicenow. oracle and others. i just think this is one, when you look at nvidia, i think we. >> sit here. >> six. >> nine. >> 12 months from now. look at this as a moment. it was a golden buying opportunity and in. >> no way. >> changed our bull thesis for where. >> this is all headed. >> nancy, let's bring you into the discussion as well. this is one of those situations where the markets have sold off, but they're not tt far away from record h

they don't need hundreds of millions of dollars in gpus and infrastructure. and that is a massive paradigm shift for the ai trade as we've known it. why the sell off today affected everything from chips to energy to mega-caps. now, the saving grace for the bulls, though, and what i hear from my network in silicon valley is that it's highly unlikely that american ai players are going to scale back their spend those billions and billions of dollars. the race only got that much more important, and the pressure to stay ahead has intensified. and the american companies, they're going to continue to throw money at this because it's existential for them. >> this is. >> all. >> just so fascinating to me. and at first i was trying to think through it and thinking, is this good or bad? i mean, isn't competition great? if we could do something more efficiently and less expensive, isn't that good? maybe not so great for america if it was done by this chinese company. when we're worried about all sorts of other ripple effects there. and that's obviously part of a bigger d

makes the really high profit, the blackwells, the $40,000 gpus that they have 90% margins.eeing those maybe not being necessary anymore. so now they're going to be questioning. on the capex spend. and i think if you think about the model itself, if you can build a model for 2%, typical model training is over $100 million. deep sea did it for under $6 million. so that in itself is fascinating to me. it's going to make stock pressure on all those tech sectors. >> all right. so jeff, we got to get going. i need a very quick answer. does this change your view when it comes to nvidia broadcom. any of these chip names that are under deep pressure. and again part of a cyclical part of the market as well. >> nvidia. absolutely. yes 90% margins are going way down. competition is good frank. >> all right jeff we're going to leave it there. your pick for us today. oracle not too concerned about the selling today. you have a lot of ideas. and that the tiktok potential could be a boost to the stock. again jeff kilburg good to see you. we want to take one more quick look at futures befor

we don't expect leading companies would move away from more advanced gpus.t 6%. cantor fitzgerald reiterating it's overweight on nvidia, calling the deepseek announcement bullish for the industry. the argument being as efficiency goes up, the man is not going anywhere. lisa: it will actually increase demand if companies can figure out how to adapt ai models to make themselves more efficient. jonathan: nasdaq futures well obsession lows, down by 5% at 6:00 eastern time. u.s. tech leaders under pressure ahead of earnings as deepseek's latest model boosts cost effectiveness while running unless advanced chips. angelo zino is with us now for more. meta says it will spend $60 billion on capex and then deepseek says maybe don't need to spend that much. is this good news or bad news for the likes of meta? angelo: an with a couple of days will do. as far as what this means specifically for meta, i don't think it necessarily changes of the story very much to be honest with you. the company announced their capex plan on friday, 60, 60 5 billion. they will reiterate th

he says with work with the cpu instead of a gpu. that is an intelligence agent and it is too slow. what you have this talk, it is under the crawl at the moment. are gti, this is a two dollars stock from a few months ago. they did a gigantic 15 million share offering. they needed the money. they had flat revenues for a long time. i don't think people are looking at these. 400,000 in revenue is a big loss. the etf has almost no quantum computing that has been bid up. i'm not saying that these are fraudulent, because willow, a project of google is pretty good. amazon was working on a program to help developers. i will use this in the sense that these companies, none of the companies have much in term of revenues. they are all losing a lot of money. it was the type of thing you see at the beginning of the year. we are seeing it in nuclear, and another area. it has been washed out before we can go higher. >> is like we did not point out any number of times as the stocks made the air moves up, about the quality of the companies themselves. i think we made reference, i made reference to t

what you think the regulatory impact will be, let alone whether inferencing will be the future for gpu demand? >> the expectation before the inauguration was this administration would double down on this type of crackdown of chips and the flow to china and we are getting a different aspect of it, so i think we will have to see how much attention the new administration is placing on this issue, but the expectation is they are not going to let up. >> how much is this complicating their efforts to get their customers to keep buying those premium ships like blackwell and the next generations? ian: it creates a massive question. if a chinese start up can find tune the models using three year old technology on the cheap, why can't you? if you are microsoft, if you are running these systems, why would you not do it cheaper if you can? that has implications for what nvidia is trying to sell. caroline: so far, the statement is all we get. let's talk about how the tech markets, roiled by china's sheep ai model. here is what some guests had to say. >> i do not think it is about china versus u.s..

v3 was trained on fairly high end nvidia gpu's and they had tens and potentially hundreds of millionst into it, work on research algorithms and they say that in the note. they do not include that. that was just for that model. they took the v3 model ended reinforcement learning to develop the reasoning model. they did not actually disclose what they did to train the reasoning model. so i do not know like how efficient it was to do that. i do know that thev3 model was efficient and there were some reasons for that. in the paragraph you just read, the models are actually really good and i do not want to discount that they are fantastic. the types of things that they did were not miracles or unknown to the rest of the community. and none of this was secret. their implementation was clever because they are under constraints due to export controls. they have to work well and efficiently. they have a bunch of really crack engineers that did a good job. the whole thing, the 5 million number just caused a complete and total panic which i do think was overblown. we saw that across the board an

show me nvidia, elon musk tweeted and deep seek has more access than gpus than originally claimed. that's an interesting idea. staying on deep seek running out of competitive ai model. what does this mean for china's influenced model and going with us for the full story. that's coming up next after this. ♪ ♪ chocolate fundraiser. with the chase mobile app, things move a little more smoothly. ♪ deposit checks easily and send money quickly. ♪ that's convenience from chase. force factor prostate advanced reduces nighttime bathroom trips, helps fully empty your bladder, and promotes a normal prostate size. why? force factor prostate advanced contains clinically studied saw palmetto, beta-sitosterol, and cranberry extract. rush to walmart and find force factor prostate advanced. all the buzz around bitcoin isn't just talk anymore crypto has gone mainstream. at itrustcapital, you can buy and sell cryptocurrencies 24/7 with the tax benefits of an ira. that's right, with an itrustcapital ira, you can defer taxes until retirement or choose a roth ira for tax-free withdrawals later. setting

the chip maker has under performed most of the peers in areas including data center gpus.nstein ing they are rmo fisher. the bio-sector could face headwinds. piper sandler moving nike to overweight. the urgency to clean up the marketplace should translate to a more visible recovery. >>> coming up on "worldwide exchange," we have the one word that every investor has to hear today and the stock pick every investor needs to know. >>> plus, a possible player in the bio-bid for the parent company sending shares higher. seven & holdings up 5%. we will have details on the high profile asset manager may be throwing his hat in the ring. that's coming up. stay with us. >>> welcome back to "worldwide exchange." time for the big money movers. taiwan semiconductor fourth quarter revenue beating the forecast as it beats the strong demand for all things artificial intelligence. shares up 1.5% now. revenues rising 4%. the company will report next week. >>> tesla with the model y in join an today. it has a longer range of 4590 miles per charge and starting price of $36,000 according to the

and it came out this weekend with the number of gpu purpose with 18 friendly nations. we are trying to get as many advantages as possible and i admit it is okay. this 200 page set up rules is the hallmark of the biden administration's treating our businesses like the enemy. trying to work with all the friendly governments and not just 18. to create sovereign a.i. . he knows america can control everything and that's how powerful we are. and with other nations, he does not want to be the ugly american who dictates things the way we did during the cold war. now the government has effectively ended it by fiat. i am no fan of the chinese government, but the u.s. stays friends. this was embarrassing, shortsighted, demeaning and admittedly fitting at every turn. unless it is clean energy, of course. of course the government's are not accurate either. i am sitting down with bristol myers and boston scientific and medtronic who have the hardest brain, heart, bone and blood- based diseases. in when the and administration target some of the biggest drugs for the so-called inflati

numbers so number one we know there are 50000h1 hundreds billions of money spent on the most recent gpurobably not accurate number one and number two what you see with the models they've completely created a new way to train models which is way cheaper and way more efficient and now anyone can take this research and apply it to their own training models, that's what you see with companies like meta and google and openai creating and putting a bunch of engineers to figure this out in now that they realized were eight days away to do this in an efficient and inexpensive way but keep in mind their line about numbers in the models not as good when you think the benchmark is very easy to cheat the numbers because they know the exact answers to provide to pass the test. when we test those and real-life scenarios they're not as good as they claim to be. maria: i tend to agree with you what is that saying fake it until you make it. i think this chip obviously has more strength than people expected is better than people expected but i'm not buying all of this type in the fact that there saying i

from nvidia's ceo and revealing why the stock's moves aren't the be all, end all for the future of the gpuhen it comes to investing, you'll love carl's way. take a left here please. driver: but there's a... carl's way is the best way. client: is it? at schwab, how i choose to invest is up to me. driver: exactly! i can invest and trade on my own... client: yes, and let them manage some investments for me too. let's move on, shall we? no can do. client: i'll get out here. where are you going?? schwab. schwab! schwab. a modern approach to wealth management. ♪♪ ♪♪ ♪♪ ♪♪ ♪♪ ♪♪ at state street, we know everyone's trying to get somewhere. ♪♪ take the next step toward your future, by investing in the s&p 500 with spy. getting there starts here. all the buzz around bitcoin by investing in the s&p 500 with spy. isn't just talk anymore crypto has gone mainstream. at itrustcapital, you can buy and sell cryptocurrencies 24/7 with the tax benefits of an ira. that's right, with an itrustcapital ira, you can defer taxes until retirement or choose a roth ira for tax-free withdrawals later. setting up an ac

these are gpu's. this is trying to keep pace with openai and microsoft and google and all of these other companies investing so heavily. it has been a huge focus for mark zuckerberg over the last couple of years and clearly will continue to be in 2025. mike: we seen investors express concern about whether there is an over-spend in ai, but they seem to be giving grace to meta. i is that? kurt: this is the most exciting technology meta could be invested in. if you say what would you rather do, have mark zuckerberg put 28 billion dollars into ai or $20 billion into, you know, the metaverse, right? obviously those things are related. a lot of the ai technology will power the metaverse. i think this is the avenue where investors are ok with overspending. even mark zuckerberg told emily chang last summer, a lot of us are going to over-spend, but that is a better outcome than waking up 10 years from now and realizing you missed the boat on this thing. mike: bloomberg's kurt wagner, thank you. it was a big we

again, the equipment to make gpus and other accelerates.ally chips from envidia and their imitators. i am surprised that envidia did not go up everyone more. the taiwan semimanagement expects this money to be well spent. they said, a higher level is always correlated with better the next year. the expenditure guide is why the group has caught fire other the past few days. capital spending from the world's largest foundry and that's the name for big manufacturing facility means money spent on the massive big ticket pieces of equipment supplied by materials, land research, and dutch asm mills of the world. the fact that most of the budget is going toward advance is also a huge positive. those are the latest and gratest and most expansive pieces of equipment hese companies sell. while the semiconductor companies had been terrific long term performers, and started out last year hot, do you know they've basically been losers for the last few months or so. only applied materials actually finish the year in positive territories. and under perform

my take away from nvidia presentation at ces is the bear argument of hardware for you by nvidia gpu andfor a long period of time it's totally and completely erroneous. every year the demand for compute, the power needed by these computers to process all the data needed for a.i. is growing annually on accidental basis. if you think about and all the data in the beginning of the history of time and the end of the year 2003 every two days. in a few years it'll be every day that every 12 hours and six hours in the computing power need to sort through that as the data gets bigger and the computation gets bigger and keep growing and growing in nvidia's chips and more powerful at the same rate. you're at a point where the latest chips you can get the same compute at a third of the cost or you can triple the amount of compute for the same cost. how do it department sit by and let the competition do everything faster, the upgrade cycle is going to be more consistent with the longer run in the most bears analysts on wall street think for nvidia in the amount of it budget and global corporate for

is fully expert control compliant here is the key inference requires significant numbers of nvidia gpuking meaning if you're thinking about it they needed nvidia chip the selloff of trying to say is a bit overdone, the destructive potential deepseek appears to have a cheaper better.pot is turning the nasdaq and ar related stocks upside down let's check the index 60747. off the lows of the session were around 749-point there is a winner and we bring this up because we want to show you a chinese company a rural mobile soaring to the moon 132% after the big data solution platform revealed the a.i. agent is integrated with deepseek are one language model but this is one of a few that are dodging the follow-up as investors flee one of the top venture capitalist in the country is not fleeing m13 cofounder and partner carter ream the firm was named a top four performing venture-capital firm at the last decade by dow jones he joins me in a fox business exclusive, wall street is losing his mind over deepseek, are you. >> not at all were in the early innings of the a.i. revolution and this is li

they're hiring engineers to work on more of that type of chip, different than a gpu.omplicated, we will get there eventually. >> nvidia was down 3%, you got me there. why am i not surprised to learn that the moderna is experiencing a lack or a decline in vaccine demand. they lowered their 2025 sales guidance by a billion dollars they say they only expect one in half, to a half-billion dollars this year previously expected to have, through half-billion dollars, low demand for the rsv vaccine and also the covid vaccine saying that they're not seeing the type of revenue that they traditionally got from the vaccines a few years ago. >> i'm not surprised what about you. >> i believe when the stock was $160 they didn't go by other companies, they had to product why didn't they diversify with inexpensive stock. >> a question, down 20% to $33 a share this morning, that's quite a comeback. tell me about bitcoin, last time we checked, 90300. >> rp quiz 108,000, as crypto falls the crypto related stocks fall with it, take a look at coinbase, riot, bit farms, micro chatterjee, th

any company that buys the platform of gpus and software can make a fortune. that's important. we keep hearing that amazon wants to compete with nvidia. something i don't believe. customers want cheaper chips. amazon is trying to make them. if jensen gives his customers a four times return on investment, that's what he is going to talk about. shows how that on monday night actual cases of how that can be accomplished i doubt amazon or anyone else could compete with that and maybe nvidia actually goes to a new all-time high. now, we get a job opening -- on tuesday, called the jolts. you know what? it might give us clues about friday's employment number. i have been mulling over the numbers and thinking about how president-elect might reverse the high levels of inflation under un-biden administration. if you close the border, you have to prayer we hear about robots at ces who skoent have enough people it in the country to do mass deportations without jacking up rages. robots may be the only hope. wednesday, interesting, very meaningful earnings reports. albertsons, yes, this groc

the taxidriver, the catchall, doesn't really care whether it's nvidia or broadcom, it's chips, it's gpu's, it's the story. you play that through nvidia, broadcom, is largely irrelevant. that is what is driving some of the other elements i mentioned. >> do you care that apple is down 3% and what do you can make of the fact that this company which -- was the best match had some of the most superlative stats in the market. >> listen, apple is a powerhouse. obviously -- not obviously cash within the context of those names has some of the least impressive growth rates. although they are looking a little better now called low double digits, but does it worry you that it's down 3%? i mean, a friend of mine likes to say sometimes stocks go up, sometimes stocks go down. >> that friend is a genius, you should hang out with that friend more often. we are leading the show with that. it's not wrong. it's not wrong. sometimes they -- by the way, great year, sit tight, come up with more ways to dunk on barry now. let's blasted out to rick santelli in chicago with the latest read on the market and why i

systems and i talked about materials simulation, but we are so far ahead of what massively parallel gpu do in an area that i don't even do that as competitive. when it comes to optimization, there are attempts to use it to solve optimization problems. it's not necessarily their sweet spot and that is an area where i think d-wave and quantum computers excel. that is why we are working with companies today to support their business operations. >> just to be clear, you are saying that it does play in the same space that jensen is talking about? i know that there is different approaches to computing and the gate model. i don't want to get into that. you are saying that huang has it all wrong? >> what i'm saying is that huang does not understand what you just said. there are different approaches to quantum. the two primary approaches. while his comments may not be totally off base for model computers, there are 100% off base for quantum computers. >> we watched it back. from dead wrong to not totally correct. you walked it back a little bit. >> hold on. when it comes to quantum computing, it

tools are already used to do things like design and engineer the integrated circuits for cpu's, for gpu's on the hardware side of the equation, they are used and employed for robust image and signal processing. the analytics for the missile warning sensors and capabilities, the integrated warning attack and assessment, certified systems all use ai to be able to push data and analyze it in a way that is actually usable to the forces that need it. modern space-based sensing systems already employ ai tools and techniques, their entire stack of their models from everything from the physical layer all the way up to the application layer and in this way, ai tools and techniques are already employed in the n c3 system. the second argument is that ai tools and techniques can help to expand the base for human decision makers. there's no doubt when you're talking about deployment of nuclear weapons these are questions of existential humanity and deserve every second that they can to be able to postulate, not only the consequence of an incoming strike, but also what the counter response is. that fix

raskin asked specifically about shipments for blackwell, as well as hopper, the previous iteration of gpusacey wanted to know if near-term guidance was going to change. jensen huang said, quote, we're not changing our guidance, i thought we did a pretty good job describing it. the ceo you can see on the left-hand side of the screen, she also said that both blackwell and hopper are shipping this actual quarter, which ends at the end of january. the total of both categories is growing and she said, quote, they're shipping several billions of dollars, but she wouldn't provide an actual number. and she also said, we'll probably do a little bit more. so that question was in regards to guidance and how much they're going to ship in this current quarter, because they said they were shipping more than they previously anticipated on the last earnings call in november. and then he did get asked a question about why were they using media tech to work with media tech, a taiwanese company. jensen huang said they have no trouble partnering with other companies and other people. and so the conversation h

the gpu is just a beast. 92 billion transistor toes. concern -- transistors. three times higher than the last generation. impossible without or artificial intelligence. liz: 92 billion transistors? i mean, they took it to a new level last night after touching a new record intraday high this morning of about $153.13, nvidia shares, as you see, are selling off a, worst perform on the dow and the knacks damage 100 -- nasdaq, down to $141. but yesterday, remember, they closed at a record. will it be tech or another sector that will emerge to lead the markets? let's get right to the floor show and joining me now, interactive brokers' chief strategist pete soz nick and ryan dietrich. steve, you know, you called it yesterday in your notes when you predicted that nvidia and the keynote would be a sell the news event. but is that anything more than just a one-day situation that we're seeing, in your opinion? if. >> good afternoon, liz. yeah, i think as of now i that's all it is. i've got a lot of questions today. you know, why are we down so much, why are the techs dow

and the gpus for the laptop. i was wondering for the laptop, that didn't seem like it should be a growth area, i mean, who wants to be in the pc market at this point unless it's a corporate pc. >> well, i guess the argument he would make is they're creating these personal super computers, so that researchers, students can do everything, you know, in an office, in a research facility or a university, for the actual laptops, he did -- and this was an interesting line, he said, i'm going to have to wait to tell you when he was asked if he was going to be entering or expanding within the pc market. so, that leads me to believe that maybe he is going to announce something at gtc in march, or in taiwan. it is a market that won't move the needle for them that much, considering data centers contributes 88% of total revenues, gaming is about 10%. you know, pcs, we know the refresh cycle has been very slow. look at amd, look at intel. it's a good point. >> the announcements were, i think, as bullish and positive in tone as

they're partnering up with micron where micron's making the memory for the gpus.eally like what you heard yesterday. stuart: i want to hear about uber partnering with nvidia because that, to me, is a very big deal. i've got a sliver of uber. >> yes. their partnering to to make autonomous driving a little bit better. uber has said they're going to be pairing the data they're getting from uber's trips with the nvidia platform to help make sure the a.i. models are working more efficiently. basically, western get better autonomous vehicles on the road given the data we're getting from these two companies. stuart: we've got all kinds of changes the free speech announced this morning by meta. including they've ended their fact-checking program. that's a very big deal. is that effective immediately? >> we think so. because with the new program that they're laying out which they're calling community notes, they say it will be rolled out in the next few months. we don't have a clear timeline yet on that new system. remember last year i sat down with joel kaplan, longtime r

we are going to teach you how to take a couple of steps down and actually manipulate a gpu, how you create applications that are built on some of the platforms from my partners here on stage. all of our curriculum is masters ph.d. level so we need partners, we need legislators, to point us to places where there are students that need to be-- that need the opportunity to see us, invite us into rooms so that we can actually have conversations that inspire the next wave of students. i was telling desmond when i missed the panel call because i am horrible at calendaring-- >> there's an app for that. >> i use -- don't mess with me, i don't pay attention to my calendar. [laughter] >> i'm just saying. and so, long story short, right? i'm at nvidia, louis, you're going to work for hbcu's, great wanting to do that for years and now having to explain to nvidia, a lot of the engineering students that i'm dealing with are not black. a lot of the teachers that i'm dealing with faculty at hbcu's aren't black. so i need to-- i need help, really, transitioning black students into the careers that we repre