Our DeepSeek stock trades - winners and losers

Today on Dumb Money, picking winners and losers in the wake of DeepSeek.

[@WingsandGills]

Everyone's fixated on LLMs. Commodities. Building blocks. Small potatoes compared to what's coming in AI. Real-world applications: self-driving, automation, robotics, biology, physics, etc. For that we need serious compute. We also need more compute if we intend to reach AGI/ASI. In what world do the big players slow down? Also, NVDA's more than GPUs used for training. It's CUDA software. It's Omniverse. It's graphics. It's an unrivalled ecosystem. Hard to imagine they won't be vital to every major project and advancement in the foreseeable future

[@etiennebreton7925]

Please adjust your 3 volumes equally, it's terrible. Thanks

[@DaleWilliams.153]

Agreed. Guy on left mic is WAY to hot.

[@sciencelabvideosl7558]

I wonder if Jordan had his own channel how much he would talk. He probably has a lot of insights but stays quiet because he doesn’t want to Interrupt the flow or the other 2 are dominant. The few times he has talked , I’ve heard gems. Even if they are just 5-10 minute videos.

[@BM_100]

Yes let's hear more of Jordan

[@tomdineen7]

Nice job guys! Interesting conversation and no egos which is important

[@gmiranda01]

Jordan is the first one to get hardware vs software with CUDA issue. Cudos!

[@dianarabbanii2]

I am trying to avoid making any new buys at this point in other not to get sucked into a bear market trap. It's tough making money in stocks when institutional investors are the driving force behind the selling.. although I read an article of people that grossed profits up to $150k during this crash, what are the best stocks to buy now or put on a watchlist?

[@LejlaGöransson]

Invest in S&P 500 ETF, for as long as possible. Do it as often as you can. Try not to withdraw this money and let compounding do its work. Prioritize patience and a long-term perspective most importantly consider financial advisory for informed buying and selling decisions.

[@angeladeem2988]

A lot of folks downplay the role of advlsors until being burnt by their own emotions. I remember couple summers back, after my lengthy divorce, I needed a good boost to help my business stay afloat, hence I researched for licensed advisors and came across someone of utmost qualifications. She's helped grow my reserve notwithstanding inflation, from $275k to $850k.

[@DanielOrstein]

I'm intrigued by this. I've searched for financial advisors online but it's kind of hard to get in touch with one. Okay if I ask you for a recommendation?

[@angeladeem2988]

Just gotta give props to Caryl Ann Adelman , my CFA, she's the real deal in the finance game. Dive into her background, this lady's a treasure trove of experience and knowledge for anyone navigating the financial jungle.

[@SuperRockcore]

Good call with ELF. Just reviewed your SMCI video. Looking forward to the next one..

[@Car-guy307]

Deepseek used 8 bit calculations instead of 32 bit, this saved them huge memory and processing requirements 😮

[@richardciotti667]

Besides the initial shock. DeepSeek doesn’t change a thing to do with the demand or price of hardware. If the race is still for the most advanced AI, once everyone adapts the new technology it’s back to hardware. This is absolutely normal.

[@Alakay77]

Chris, with respect to your knowledge and ideas, can you please let Jordan finish a sentence! You ask him a question just to interrupt him and assert your own ideas instead.

[@VoltageLP]

Fun Fact DeepSeek does answer questions about Tiananmen Square, but only if you ask them in russian. It won't answer if you ask it in English

[@banana19934]

Very telling

[@Mediassshow]

Wow

[@Vev79]

На русском она считает себя YandexGPT. На английском OpenAI. Если запустить локальной или через арену AI, то и на английском не зацензурена.

[@cryptlol4736]

Spoiler alert- long nvda is the best trade.

[@tinman1071]

exciting show thank you

[@kg3436]

Nvidia definitely wasn't blindsided- Jensen talked about time scaling and lowering the cost of compute at his CES presentation before the deepseek reveal.

[@CC-kl4nh]

I want to learn from Chris and grow my money before I die.

[@marcpeterson8530]

Dario Amodei just said in an interview that the current RL optimization’s are helping the models greatly, and do create downward pressures on the cost of enhancing the current models. But given RL will also create tons of synthetic data, model makers will have to go for those hundreds of thousands to million size clusters to make the next jump. So in the interim, these cost saving techniques are making up for the lack of compute, but there is no end in sight for the need of bigger and bigger pre-training runs to achieve the types of AI (AGI, ASI) everyone is vying for. Bullish NVDA

[@elianafeng1926]

What are your thoughts on Tesla robots vs. Chinese competitors like unitree?

[@eugeniustheodidactus8890]

Grok will be incorporated into both Optimus and FSD.

[@jwinstonh1926]

The smartest folks here would assume that what is presented by Deep Seek is true. It can only motivate you to do better. Keep this in mind, the time spent dragging down someone else is time not spent on uplifting yourself.

[@ShawNshawN]

China is notorious for rolling out fake tech jumps in tech

[@elborro444]

I think FSD might be an exception to the inference side efficiency improvement, if it were to compromise on security. Every little bit counts when it comes to safety. Anyway i have no idea for sure but just a thought

[@cw5335]

It’s a situation of competition between Chinese Chinese vs American Chinese. BTW the CEOs of NVDA, AMD, AVGO are all Chinese 😊

[@stickeroonie]

Just remember at every other inflection point in history you'd be a genius if bought at the highs. Meta at $80 people talking about 60$ meta or $100 meta. All a moot point.

[@deenobrown1]

Why hasn’t anybody just started with the simple math? Even if deep seek is using only the 2000 H 800 that they said they have, the ones they are allowed to have, it would cost way more than $5 million. Each one of those GPUs in America cost between $20-$40,000. There are reports that the Chinese were paying $70,000 for each H 800. That would be between 40 million and $140 million just for the GPU. That’s if they only used the 2000 that they said they used. I don’t understand why people haven’t done the simple math. $5 million divided by 2000 GPUs is only $2500 per GPU. There are not any GPUs in the world that only cost $2500.

[@ShawNshawN]

Facts

[@MeditationMindless]

PLEASE STOP SAYING NIVIDIA WHAT IS WRONG WITH YOU DAVE??

[@eugeniustheodidactus8890]

*Mathew Sharpe* Tesla has raised prices in many markets and they typically only do this to throttle demand. Record sales in China. The refreshed vehicles are selling very well with Tesla's newer, cheaper models coming out very soon. ( EV incentives in EU and China are back. ) ENERGY should more than make up for the loss in revenue from slightly lower auto sales for 2024.

[@erichharrington1213]

The US TSC fab is for Intel level chips.

[@73mcox]

Perplexity for search is the best on the market

[@Will-be-free]

I'm using it, but it's annoying. When you google something, then it sometimes doesn't find what you are looking for. When you search with perplexity, it will tell me that the thing I'm looking for doesn't exist, just because it can't find it. (I was searching for a quote by a politician, and perplexity claimed that she had never said anything like that, even though I know I've read it before.)

[@JC-nt3qt]

What do you think about XIFR stock?

[@Mr.Mindset]

One small edit: OpenAI is not the biggest Nvidia client!

[@LiLSTRONG-w9p]

16:41 ..actually had me nervously thinking

[@gray-jones]

Jordan looks exactly like Andy Roddick, my favorite tennis player ever

[@cal4865]

Buying $NVDA 3/21 Calls buying all weakness!!! Guaranteed Winner 🥇

[@z5m2]

Deepseek news was out for a while. Unlikely they timed it well enough to just short b4 the weekend

[@kenplant91]

All of the spending on chips is with an end goal in mind, to create the ultimate a.i. The spending on chips will not slow until we get there. The gas just got cheaper, that's all. imo

[@ceciliaferraro6360]

Thanks

[@joshuab2450]

And just like that Deep Seek drops another incredible Multi Modal Model....LOL. Long China!!!

[@ericb9328]

How come no one is talking about power consumption with all this computing power that is going to be needed? Power plants are still shutting down in the US with very little development going on.

[@N34R4T0M4T]

I am new to this podcast you guys are on the money BUT you guys need to let each other speak and not interrupt.

[@Godsavethecrumpets]

The US will certainly implement more export controls. But conversely I think it could be an opportunity for nvidia to increase gross margin. Decreased supply equals higher price?

[@Will-be-free]

Wouldn't stricter export controls just mean less demand from China?

[@BobCMooremfp]

Maybe NVDA last earnings was a ‘tell’ on what was coming…

[@Anthony-bo8ry]

Would software companies become the beneficiaries because the access to the models becomes cheaper for them to implement?

[@erichharrington1213]

Tesla networked 100K GPUs together without NVidia's hardware.

[@khalidkamal5629]

As the name implies Open AI was intended originally to be open technology for everyone and then they went into a for-profit model and got away from open source. China on the other hand kept up with open source and was able to make the AI they have better simply because the open source community was helping to refine the code. If that AI bubble gets burst it's only out of pure greed that it happened. I don't have much enthusiasm for the proposed AI infrastructure because that is only code for creating the infrastructure for a surveillance state. It all sounds good until you realize you don't have any control over that infrastructure and you will be under the all-seeing eye.

[@Patokefe]

Will Altman be replaced in 2 weeks or 2 months? And then what?

[@erichharrington1213]

NO!!! They still need the biggest Datacenters. We are FAR from the endpoint on AI, and these big datacenters are still necessary. All that happened is the industry got a 1 yr bump. All the big guys would take a year or two to get to where we are now.

[@IncognitoMan9]

Guy on the right needs to learn to not interrupt others.

[@TeslaCiviliz8n]

1:06:35 Honest question. If the majority of Tesla's valuation is based on FSD, then why does the stock never move more than 1% on bombshell announcements about FSD? Were you paying attention to the stock after Robotaxi Day? It went down a little bit.

[@TeslaCiviliz8n]

1:13:30 Every time Elon has mentioned priorizing FSD over vehicle profits, the stock has gone DOWN. It's been visible during earnings calls. You can also just hear it from Wall Street investors anyway. Before last year, they were not accounting for it at all. Now some of them are accounting for it a little bit.

[@stickeroonie]

Long term this guy coulda held his nvidia shares and probably smell like a rose

[@cyril2702]

Apple needs to release a 10k$ double m4 ultra mac studio, it’d sell like cupcakes

[@jasonng1299]

Once agentic AI is at a useful level, search is pretty much done.

[@LUCKY_LARRY]

What are the potential implications for AI robots? Will US companies dump capital into a black hole, and the Chinese will roll one out of the closet that's been developed for peanuts and is just a wee bit better than Tesla's?

[@Christian-ry3ol]

The ultimate winner is Meta. Not Amazon. It benefits Meta's business in every way.

[@jekatskin3860]

Crasy it was 80 not so long ago...

[@andrewvercillo7584]

What about Tesla as a winner. No one thought a Model Y had enough compute to handle FSD! Now this proves Elon again that we can operate FSD on our cars! Hardware 3 cars will run FSD

[@khalidkamal5629]

I am sure that Elon is jumping from the joy, you know why.

[@erichharrington1213]

The big guys like Tesla will knock off and learn from Deep Seek and make it better.

[@peace_4joy]

All 3 tomorrow

[@ChaceBonanno]

Bruh it’s pronounced N-Vidia, not nvidia

[@MeditationMindless]

Nividia

[@stickeroonie]

Potato Potahto

[@BobbyWF]

What is the parallel with TikTok? Can it be bought or banned?

[@kikimo57]

AI will be everywhere and chips will be needed even though they're not needed now. The growth will be exponetial.

[@upcomingexpat]

I'm not going to knock, DeepSeek. I hope some of it's true. I can't believe close to $6 million was spent on this. The amount on Nvidia chips (that they weren't supposed to have wasn't $6 million.

[@angelodorsett1]

The focus should not be on the LLM.

[@mordechailebovits6338]

Encoded in the large, highly evolved sensory and motor portions of the human brain is a billion years of experience about the nature of the world and how to survive in it. The deliberate process we call reasoning is, I believe, the thinnest veneer of human thought, effective only because it is supported by this much older and much more powerful, though usually unconscious, sensorimotor knowledge. We are all prodigious olympians in perceptual and motor areas, so good that we make the difficult look easy. Abstract thought, though, is a new trick, perhaps less than 100 thousand years old. We have not yet mastered it. It is not all that intrinsically difficult; it just seems so when we do it.[7] A compact way to express this argument would be: We should expect the difficulty of reverse-engineering any human skill to be roughly proportional to the amount of time that skill has been evolving in animals. The oldest human skills are largely unconscious and so appear to us to be effortless. Therefore, we should expect skills that appear effortless to be difficult to reverse-engineer, but skills that require effort may not necessarily be difficult to engineer at all

[@mordechailebovits6338]

Wait till robotics and motor skills come into play

[@AndyNick211]

Dave, love the show but you gotta learn how to pronounce NVIDIA correctly 😂

[@AdaAng-i1t]

A closed economy / industry thru protectionism breeds complecancy n crushes innovation as well as creativity. On the other hand, an open mkt / industry with healthy competition drives innovation n made america #1 in the world esp for tech n inventions. that is what a laissez-faire economy is all about and which resulted in the America we have today. However if we continue to go down the current protectionism road and with the rest of the world retaliating, it will be a very different world

[@Will-be-free]

The monetization problems with software were there already before last week. I don't really see how that has changed. Yes, someone may have come up with a better way to run the software, but that doesn't mean that you wont get even better results if you run it on better hardware. Chat bots are not the end goal for anyone. Whatever nvidias customers are working towards, they just came a step closer to their goals. What's the reason these AI companies don't just empty their cash balances and buy all the chips all at once? Well maybe today's chips aren't strong enough to reach the goal and they may have to wait for next years hardware. If what you can make with current hardware has just been pushed forward a couple of years, then it would make sense for companies to buy more now rather than wait, because chances are much greater now that what you can buy now will be enough for a monetizable product. I'll bet anything that any serious contender in the AI space will invest heavily now, or they will be lagging far behind. If you can do more with the same hardware, it just speeds up the race.

[@BillSmith-fx7xx]

I'm thinking the airlines are gonna benefit. We're all gonna be traveling with our R2D2. Twice the number of tickets ! :- )

[@elborro444]

Jordan’s take on post deepseek is kinda baffling, like expand on those one word answers.

[@peterzz4266]

Imagine the money made if they bought weekly puts otm and released thier news

[@CLC077]

Why are they taking random hypotheticals …

[@peace_4joy]

Just in Trump is deporting anyone that say’s Tesla is only a car company 🤣

[@tonychow785]

Will there be a Deepseek moment for Tesla ? Will cheap Chinese knock off humanoid robots change your TSLA thesis?

[@topcolas7075]

Chinese simply focused on software rather than hardware..

[@georgeortiz4169]

Can you three please start using the same mic? Or at least tune the input volume equally between? The volume delta between Dave and Jordan is too wide.

[@Kcpmaui]

I’m a little worried about the tariffs trump’s claiming to put on Taiwan this might be the black swan for semis

[@jsssm]

Lol chris thinking an LLM would somehow help fsd... Come on.

[@DAQtrader]

Daves mic is too loud in proportion

[@mnab7616]

Chris is great! But he talks too much and doesn’t give air time to others, would really like to hear what others say, and he cuts them off that’s a bit put offish.

[@VoltageLP]

Home AI builds would be built on either cheap server hardware or premium quality consumer HW. Corsair is neither, still dead.

[@DerivCapital]

cheap knockoffs are never better than the name brand...THIS is not that ...THIS IS the LINUX of AI so i think you better put some respect on Deepseek's name

[@jonknight1408]

PLEASE for the love of god Dave: Nvidia is said in-vidia or en-vidia NOT nuh-vidia 🙏🏻🙏🏻🙏🏻

[@peace_4joy]

I posted China figured out how to short the US market

[@peace_4joy]

Wow, you just gave a super bullish statement for Robin Hood 24 hour market

[@Zugzuggers]

XRP

[@BobCMooremfp]

Last TSLA earnings my strangle play paid off Big. They did have their biggest earnings gain yet, so that helped. Elon has to come out guns blazing tomorrow.

[@mshparber]

They are fking liars! How people are buying this $6M???

[@Car-guy307]

Even $60m is cheap

[@Mediassshow]

Well double it... triple it STILL SIGNIFICANTLY cheaper

[@deenobrown1]

they said they use 2000 of the H 800s. Each H 800 cost between 30 to 40,000 in America and there are reports that Chinese pay up to 70,000 for each one. That would cost between 40 to 140,000,000 just in GPUs. It’s really simple math. If they only spent 5 million that would mean each GPU only cost 2500.

[@ondramlcek3174]

Plus APPL of course. Not even mentioning their 0 innovation approach over the last so many years... The company is dead from the inside.

[@acs2777]

Alibaba is out with their AI 🤖 chat .., 😊😅

[@ololoberriesful]

IT IS NOT NUH-VIDIA!!!

[@disastrousemouse]

You guys are going to lose a lot of money, aren’t you?

[@fnad3999]

Google hardware store near me Google roofing supply near me Google best burrito near me

[@VetejDEDU]

Every one blind to shopify 🎉😂ty🎈🎁100/gain

[@donjohnson4368]

Sound is horrible - stopped watching after 5min unfortunately

[@mcontaldi3654]

DAVE, it's pronounced "en - vidia" !!!!! :)

[@erichharrington1213]

JFC! This is all nonsense. If Deep seek can do as much of 5% the Datacenter, what will it do on 100K GPU's. MORE. What were doing right now isn't the goal. It's General AI and they are far away. So All this does is allow the big players to copy the good parts and use the innovation to their advantage. The moat was never the AI softwre, it was the Datacenters and what that power allows them to do in terms of training. The Chinese model used US AI's learning to train it. It didn't do it from scratch. As usual, the chinese copied and improved in some ways. The fficiency is just NOPT a game changer, except it does open AI to many more companies. So the big guys counting on the datacenter moat, like Open AI, may be hit big, but it shouldn't do shit to NVidia, becuase again, Deep Seek is Open source and if it does that well on less, imagine what it does with 20 times MORE!!!

[@REDCLAYHOMESTEAD]

23:59

[@REDCLAYHOMESTEAD]

45:00

[@SomeInfo-ib3wz]

Yeah...dumb money.

[@ele3230]

I have a bridge for sale, anyone interested , excellent views of lower manhatten and Brooklyn , who’s interested

[@thomasthetankengin7722]

Stop saying Nuhvidia Dave! Ffs

[@73mcox]

Micron Technology is going to be the big winner in FSD in tomorrow earnings