Jlagreen

And yet, Nvidia innovates in gaming with SW 10x more than anyone. Strange side business it is which is also worth billions by the way.

Did you know that Nvidia has a gaming cloud running which might become the largest in the world at some time?

In 10-20 years, Nvidia might make more revenue from gaming cloud than they do today with gaming HW.

Because Nvidia focuses on being a partner in each industry you mention.

See it that way, if you have an OS/SW for all the industries you mention then who is your competitor? Not the participants in that industries. Nvidia can partner with any automotive company but won't compete with any of them as long as they don't build cars. But imagine the potential of every self driving car being build using Nvidia AI?

Think about the potential of every robot build using Nvidia AI?

Think about the potential of any AI Service using Nvidia AI?

See, Nvidia isn't directly competing in the enduser market but instead focuses on the B2B. Nvidia can also create many different revenue streams from 1 customer.

For example an automotive customer: - Nvidia HW in car for AI - Nvidia data center on-prem/cloud for DriveSim in car - Nvidia Omniverse for car design and manufacturing simulation - Nvidia Isaac for robotics/logistics in manufacturing plant - Nvidia Cosmos+GR00T for robots inside the plant - Nvidia edge devices inside any robot in the plant - Nvidia NeMo data center on-prem/cloud for AI models / LLMs for internal use

And what will be the advantage? Nvidia can actually make it more and more seamless to operate between all Nvidia solutions. For example, you can do an update to your robots in Cosmos, simulate it in Omniverse and with 1 click update your Nvidia driven real robots. The alternative is that you have 3 solutions from 3 different vendors with no interface between them.

People have no idea, what Nvidia is actually creating. Nvidia has more SW engineers and even Nvidia employees call Nvidia an AI SW company. They publish so many libs and lots of other SW stuff that it's sometimes hard to keep up. Just look at all the RTX goodies for gamers which Nvidia is developing. And they are all free, well except that you need Nvidia HW for it. The same model, Nvidia will apply to ALL industries in the world. And here they discuss about CSPs being an issue for Nvidia while Jensen focuses to build a Mega Corp. which potential TAM is in every industry in the world :)

One of the most important contributors to QC is Nvidia because they try to help with GPUs in creating SW for QC and simulations. Nvidia has cuda quantum for years now.

Yes, but the cloud customers "who finance TPUs" have NO INTEREST in TPUs and in Nvidia GPUs instead.

How does Google pay for TPUs internally? By Google Search and Google Cloud of course. Google Search uses TPUs, Google Cloud however has way more non-TPUs instances.

What people forget, nobody wants to switch from CUDA dependency to SW dependency on Google/AWS/Azure. CUDA at least allows me to use it in consumer, in pro HW, in cloud and AND in on-prem data center.

I'm really looking forward to Fortune 500 companies sending all their internal company data to Google to structure it to train custom AI models. Yeah, that will never happen. What happens instead is that Fortune 500 companies will build up AI expertise to build their own custom AI model and they will think hard if they want the training AI compute internally or on a cloud. Nvidia has a huge business of building data centers on-premises which people totally oversee. NO CSP will ever compete there because it's against their primary business model. A Reliance India contract from 2023 alone is a delivery of 2 million GPUs in a few years. That's probably more than Nvidia's last year's total revenue and that is 1 large corp in India only.

Yes, we can see how Big Tech and Apple's huge margins get compressed all the time, they never expand. They are going to zero soon it seems :(

You can't say that in general because it also depends on the moat.

Apple has 75% profitshare in the smartphone market not because they have the best smartphone but because they sell iOS. This is why Apple can charge much higher margins on iPhones then any other smartphone competitor. Competitors use Android and are basically exchangable so their HW is commodity more or less and margins are much lower.

The same will happen with Nvidia. Nvidia will offer complete data center solutions and many other SW/HW solutions for AI and accelerated computing and will charge high margins for that. HW sellers like AMD will sell only chips which will compete with commodity ASICs.

What are you talking about?

Nvidia has been a high margin and great performing business for the last decade. Nvidia had better gross margins than apple by selling gaming GPUs only 10 years ago and that's in a market where you can easily exchange the card in a PCIe slot.

From 2015 till 2022, Nvidia had several years with 50-60% revenue growth. People only look at the recent 2 years and think that Nvidia was "OK" before but I'm invested since 2016 because Nvidia started the growth turbo back in 2014/2015.

Jensen decided decades ago that Nvidia is premium and he positions the company in that position. What many don't get, Apple has only 25% unit share but 75% profit share. So Apple basically concentrates the profit of the Smartphone business. Nvidia will do the same. They had better gross margins a decade ago than AMD has today. AMD might gain unit share but will never gain Nvidia's margins because AMD is a market follower and will never be able to set pricing unlike Nvidia with premium solutions.

Jensen also made CUDA possible. Intel on the other hand killed such projects and also their first GPU project. Intel also didn't invest in OpenAI and so on. Jensen is by far the best CEO and fortunately he doesn't get crazy like Musk does.

The way I see it is that Nvidia might reacht $1 trillion in revenue before AMD reaches $100 billion in revenue. So the upside in revenue growth is higher for Nvidia.

People think that because a company has grown very large very quickly that it can't grow as much anymore. But on the other hand, there is clear evidence that Nvidia continues to dominate AMD's offerings despite the latter having a competitive product now. So the metric for Nvidia isn't Nvidia vs. AMD but the growth aspect of AI market overall.

Ah, I'm sorry you're right a misunderstood your comment.

I agree and Nvidia positions itself for exactly that. See how fast DeepSeek will come to NIM. People are already wondering how well DeepSeek will run on Digits.

Nvidia also offers distilled open models or specific own open models so is indirectly competing in that space as well. But Nvidia isn't in the LLM generator business but in the business of "infrastructure for LLM generators"

Everyone is waiting for GPT5 or another big bang. And because it takes so much people start to think that there is a wall. And there is a wall but that wall could be also compute. Blackwell will show if there is a compute wall because simply put, if a training run with large parameter set on GPT5 takes like 4 months then Blackwell might reduce that to under 1 month with the same amount of GPUs. Getting more GPUs can get that down even more. Imagine the speed up in AI frontier model research if your training times come down 4-5x from new GPU generation and another 2x from getting twice as many.

The nice part with Nvidia is also that the old GPUs don't become obsolete, OpenAI can continue using them for inferencing or even try to use combined architecture training as long as they don't go FP4.

I wouldn't be surprised that at the end of 2025 we will see things which will make DeepSeek and GPT4 as oldschool stuff simply because of the massive compute which Blackwell will deliver this year.

That was never the moat.

Nvidia's moat is controlling compute for 80-90% of workloads.

Why do you think Steam has more RTX 4090 than 4080 in their survey?

Ever since RTX 2080TI you could buy multi server consumer GPUs for ML. We sell PCs with RTX cards regularly to customer for small local AI applications.

Project Digits is the next thing. It is not only a Nvidia GPU in some PC, it's the real AI PC. The real deal and we're already considering switchting to that as a system instead of a PC since it's size is perfect for our Vision application.

Do you think Nvidia cares if you buy 1 Blackwell DC GPU, 13x Digits or 20x RTX 5090? In the end it's all the same turnover for Nvidia.

Nvidia's goal is to spread and dominate workloads worldwide and that no matter if DC, enterprise or consumer, Nvidia HW is used.

Nvidia had 65% margins 5 years ago when they primarily sold gaming GPUs.

What margins do you think Nvidia is charging for RTX which getting more and more expensive every generation?

What margins do you think Project Digits will be for Nvidia? That thing costs 50% more than a RTX card and on the RTX card Nvidia only earns on the chip which is only part of the card. Digits is a 100% Nvidia product.

Does it matter if 1 Blackwell DC GPU or 13 Digits are sold? No, Nvidia has probably an insane margin on Digits and has only one goal, total spread and integration of Nvidia HW in any AI workload. Because then Nvidia can offer SW with super margins.

Think of Nvidia Enterprise AI, Omniverse, Clara, Isaac, DriveSim and many more. If 95% of AI acclerators are from Nvidia then it's only a small step to also use Nvidia SW frameworks.

Without Nvidia, you would still wait for ChatGPT moment and not even think about AI at all.

I have been invested in Nvidia for 9 years and I have not only witnessed what Nvidia has done in ML/AI but also how the entire field evolved.

To say Nvidia is toast is like saying US is toast after Sovjets send a rocket into space.

So the focus is trying to find things where LLMs are bad instead of trying to find out where they are good and find applications for that?

That's basically like trying to embarass a IQ 180 student on emotional intelligence.

But I guess that's human nature to expect a machine to be 100x better than humanity on first try.

is llama405 a distilled model like DeepSeek or a trained frontier model? I honestly ask because I haven't researched but that's important to know before one compares.

For 2 decades we were told how Apple will have to cut their margins due to competition and so on.

Today, it's simple. Apple has 25% unit share in smartphone markets and 75% profit share. Apple makes 3x the profit of ALL OTHER smartphone vendors combined.

And this is exactly where Nvidia's goal is. The AI compute market will grow, Nvidia will lose unit market share but Nvidia will retain their profit market share. Simple as that.

And by the way, Nvidia is way ahead in SW compared to alternatives. Most here have the DIY glasses on. But enterprises and businesses have different lenses. For those not being Tech they need secure and working solution with enterprise grades. Nvidia is among the few to offer this with Enterprise AI solutions (NeMo, NIMs, etc.). Nvidia's SW moat isn't CUDA, CUDA is an API for performance and stability. Nvidia's SW moat is in the frameworks for applications for many differnt industries and of course ALL Nvidia SW will require Nvidia HW.

A company using Nvidia enterprise SW solutions and consultancy will never use anything except Nvidia HW. Nvidia has a program with >10k AI startups being supported with free consulting and HW support. Nvidia is basically grooming their next generation customers by themselves.

You have no idea, many think Nvidia is only selling some chips and that's where they are wrong. Nvidia is a brand, an ecosystem and they will continue to grow from there. See gaming, much more standards and commodity in SW than AI SW. There is no CUDA, you can swap a Nvidia card with AMD card within a minute. So let me know, how come for 2 decades that Nvidia has continously 80-95% market share?

Nvidia is already doing that.

What needed 1000k of Voltas, needed 100k of Amperes, needed 10k of Hopper, will need 1k of Blakwell.

Nvidia has increased compute by a factor of 1 million in the past decade and it's no where near enough.

Blackwell will increase training efficiency in large clusters a lot compared to Hopper and yet it's already sold out because even that won't be enough.

The good thing is:

Blackwell DC is $40k per piece and Digits is $3k per piece. So if 13x Digits are sold then it's the same turnover as a DC GPU for Nvidia. Yes, maybe lower margin but Nvidia can easily scale digits into masses compareds to Blackwell DC GPUs.

In the end, the winner is Nvidia because Nvidia doesn't care if DC GPU, Gaming GPU, Digits GPU, Jetson GPU is used for AI as long as Nvidia is used 98% of time for AI workloads. That is the world domination goal, simple as that.

And that's what Wallstreet doesn't get. Digits is 50% more turnover than the largest RTX GPU. On average gaming GPU turnover is probably around $500 per GPU. Nvidi probably sells 5 million gaming GPUs per quarter. Imagine they could reach such amounts of Digits. That would be $15b revenue and almost half of current DC revenue with Digits only.

And yet, Blackwell is sold out.

What does that tell us?

The industry is compute starved and that makes totally sense.

The tranformer model on which current LLMs are based on are 8 years old. But why took it so much time to get to the LLMs only 2 years ago?

Simple, Nvidia first had to push the compute at scale strongly. Try training GPT4 on Voltas from 2017. Good luck with that!

Current LLMs are possible thanks to the compute Nvidia has provided in the past decade. You could technically use 20 year old CPUs for LLMs but you might need to connect a billion of them.

The $5M was the cost of the training itself.

You can rent 10k H100 for 20 days with that money. Go and knock yourself out because that compute is probably higher than what DeepSeek received for that money. And that is public cloud pricing for single H100. I'm sure if you ask for 10k H100 you'll get them at half price so easily 40 days of training.

DeepSeek has fooled everyone by telling them that they need only so less money and people think that they only need to "buy" $5M worth of GPU but that's wrong. The money is the training costs of renting the GPU training hours.

Somebody had to install the 10k GPUs and that's paying $300M to Nvidia.

Your implication is that we have unlimited compute and therefore know that LLMs are stalled.

Have you considered that compute might be the reason why LLMs are stalled at the moment?

What made LLMs possible in the first place? Right, compute! Transformer Model is 8 years old, technically GPT4 could have been released 5 years ago. What stopped it? Simple, the compute being way too low.

Nvidia has improved compute by 1000x in the past 8 years but what if training GPT5 takes 6-12 months for 1 run based on what OpenAI tries to do?

What we see right now is that pre-training has reached the limits of Hopper and Big Tech is waiting for Blackwell. Blackwell will easily be 10x faster in cluster training (don't look on chip performance only) and since Big Tech intends to build 10x larger GPU clusters then they will have 100x compute systems.

Let's see then how it turns out.

The limit on training is time. If you want to make something new and improve then you should limit training time because nobody will wait 5-6 months for results anymore.

It was fine for OpenAI years ago to take months to years for new frontier models. But today the expectations are higher.

There is a reason why Blackwell is fully sold out for the year. AI research is totally starved for compute.

The best thing for Nvidia is also that while AI research companies compete with each other, they all try to get Nvidia AI HW.

That's wrong.

DeepSeek is a problem for Big Tech, not for Nvidia.

Why?

Imagine a small startup can do something better than Gemini or ChatGPT or Claude.

So it can be disruptive.

What can Big Tech do to avoid disruption? Buying every SINGLE GPU Nvidia produces! They have the money and they can use the GPUs in research.

The worst nightmare of any Tech CEO is a startup which disrupts you so you have to either be faster or you kill access to needed infrastructure for the startup. Or even better, the startup has to rent your cloud infrastructure, this way you earn money and you have an eye on what's going on.

Additionally, Hyperscalers only get 50-60% of Nvidia's supply. They all complain of being undersupplied yet they get only 60% and not 99% of Nvidia's supply. How come? Because Nvidia has a lot of other customers they like to supply to. That alone tells you how huge the demand is that Nvidia even has to delay Big Tech deliveries.

Also the demand for Nvidia didn't drop. DeepSeek isn't a frontier model. It's a distilled model therefore the moment OpenAI, Meta or the others release a new frontier model, DeepSeek will become obsolete and will have to start again to optimize.

No, Nvidia's margins won't drop at all and the proof for this is Apple.

The units of AI accelerators will explode, the market will explode.

At the end of the day, Nvidia will have 20-30% of the unit share in AI HW and 70-80% of the profit share in the AI HW market. Just like Apple makes 3x the money compared to the rest of the smartphone market.

Jensen has considered Nvidia a premium vendor for 2 decades and track record of Nvidia's margins show this.

And while Nvidia remains a high premium AI infrastructure vendor, they will also add lots of great SW frameworks to make even more profit.

Omniverse has literally no competition. That digital world simulation combines all of Nvidia's expertise (AI HW, Graphics HW, Physics HW, Networking, SW) into one huge product. And it will be a revolution because it's the first time we will be able to finally digitalize the analog world. And Nvidia will earn tons of money because Omniverse itself is licensed, it needs OVX systems (visual part) and it needs DGX systems (AI part).

Don't worry, Nvidia's margins will be totally fine. I would even expect them to be higher in 10 years than they are today. Nobody believes that but that's Jensen's goal.

There is a reason why Nvidia has always been the company with the highest P/S ratio and anyone who understands why, will see the quality management immediately.

My entry into Nvidia is 2016, my portfolio has never been red since then.

Deepseek was trained on Nvidia GPUs H800. Nvidia is still selling GPUs to China and the reason even the reduced chip performance is easily negated by scaling.

US government is stupid because they asked for certain limits on chip base first but AI uses GPU clusters. In a GPU cluster you don't have full utilization anyway so slower GPUs don't matter as much as slower networking. China still gets pretty high bandwith Nvidia HW for building large clusters for training/inferencing.

Chips from Huawei still seem to be way too unstable for the job. In training/inference stability is even more important than performance. Imagine you have a fast chip but it can't run without errors for 2 months and you training never gets done. That's DoA.

I would shoot myself if I had to wait 9 seconds for a query. I sometimes even would like to kill by browser taking seconds to open a page...

Yes, so Chinese did what they are best at, taking something working (foundational model which is free), copying it and improving it.

Nothing special to see here.

The good thing is that Chinese models are dependent on open sourced US models.

I wonder, will Meta's next models remain open source? Zuckerberg might change his mind.

Who knows, maybe even US will dictate that no US model is allowed to be open sourced lol. I mean the restrictions but US government in AI have been crazy, especially the last ones by Biden administration.

What do you think ASI will be?

We will only advance in AI if one day AI learns from itself otherwise we don't need AI. Because AI learning from itself at X times faster than humanity is the reason we want to get there in the first place.

That's why the shovel maker from back then are selling mining machines today.

Everyone here thinks Nvidia is dommed because of training efficiency.

But what has Nvidia been doing for the past decade? Correct increasing training and inferencing efficiency by magnitudes.

Try to train GPT4 on 10k of Volta, Ampere, Hopper and then Blackwell.

What has happened since then? Nvidia has increased their sales in magnitudes.

Why? Because thanks to improvement in data, in algorithms, compute efficiency ChatGPT was possible in the first place.

Imagine Nvidia wouldn't exist. When do you think the ChatGPT moment would happen on CPUs? LOL

Going back to my first sentence. Nvidia started also with small shovels which were GeForce cards with CUDA. Today Nvidia is selling huge GPU clusters (mining machines, yes pun intended ^^).

THIS is the only correct statement in all of this.

The goal for AGI and ASI MUST BE to train, inference, train, inference and so on and that all on the fly in fractions of a second from every token produced.

Now good luck calculating the compute and hard work in algorithms to get there.

Not possible? Then AGI won't ever work because how can AGI beat a human if it can't learn on the fly? Not to mention ASI lol.

The issue is it feels like we came to a stop but Hyperscalers are simply waiting for Blackwell. That's all.

Why buy 100k Hoppers if 20k Blackwell offer the same compute so then it's better to buy 100k Blackwells right?

Backwell will increase cluster scaling easily by 10x performance and if you buy 10x of them then your compute on a cluster will be 100x than before. If it takes you to wait 6-12 months for that then so be it. You will easily make up the time in the end with the speed up.

$50 billion can be afforded by WAY WAY more sites and companies so Nvidia will simply delivered to >10x more data centers. Or instead of shipping 100k GPUs to Meta, they will ship 100k GPUs to 10 different customers.

For Nvidia, it is great news because now finally, the concentration of GPUs at Hyperscalers will end and every Fortune company can finally get their local data center to train their AI Models.

Because if training AI models becomes more efficient and easier then the ones being that business are at risk so basically Big Tech. Nvidia isn't in that business but in the business of providing tools to train.

Fortunately, Big Tech can easily do something to prevent ANYONE for competing. They simply buy all available GPUs. Oh wait, haven't they been doing it for years? Excatly!

People really don't get what an arms race and market competition race is.

How do I prevent disruption? I simply buy all the tools the competition needs to disrupt me.

See, if Fortune 500 companies want to build large data centers but can't because all Hyperscalers buy the GPUs then eventually they will rent from cloud as otherwise they can't get the GPUs.