NVIDIA's Future: Scaling AI and Market Leadership Insights

NVIDIA's leadership in AI and computing relies on its integrated hardware and software solutions, driving future growth.

### NVIDIA's Dominance in AI and Computing: Insights and Future Prospects
* 💡  The key information desired from Jensen Huang is confirmation of **ongoing scaling law improvements in AI**, indicating that significant investments are leading to **advancements in technology**.
* 📈  Mark Zuckerberg's comment about future models requiring **ten times more compute** highlights the **increasing demands placed on NVIDIA**, which is crucial for its **long-term growth narrative**.
* 🤖  The transition from clusters of **less than 100,000 to a million** will enable the training of **larger and more efficient models**, making it essential to **evaluate the return on investment in AI**.
* 🏢  While some companies see **massive revenue growth** due to AI, others may only **maintain competitiveness**, drawing parallels to the widespread adoption of **Microsoft Excel**.
* 🌐  NVIDIA's potential to serve as a cloud provider is under scrutiny, with a focus on whether it will **develop its own cloud** or continue selling through **existing hyperscalers**.
* 💰  Daniel Sams Capital's substantial assets reflect the **strong demand for NVIDIA investments**, demonstrated by the launch of leveraged ETFs targeting NVIDIA's stock movements.
* 🔮  The consensus is that NVIDIA's leadership in computing will **grow over the next 5 to 10 years**, driven by increasing demands across various sectors such as **healthcare and automation**.
* ⚙️  NVIDIA's competitive edge stems from its **early recognition of market needs** and the **comprehensive software ecosystem** it has developed, making it **challenging for competitors to catch up**.

What would you want to hear from Jenson, regardless of whether he actually tackles it? What is the number one piece of information you want him to give you tonight? So I think the number one piece of information is confirmation for ongoing scaling law improvements. And so the idea is to say that you're throwing tens of billions of dollars at A.I. and AI continues to improve. I mean, we heard Mark Zuckerberg on his earnings call saying that training the next model is going to be ten times more compute-intensive. Those types of comments are very important for the long-term story of NVIDIA. And I think actually the core focus of the call is probably going to be on some sort of theme like that because that drives the growth the next one, two, three, four, five years. So sort of our scaling laws still holding true. And what's the evolution of AI within that? So it's quite a long-term picture you

want, Daniel. And what's interesting is everyone suddenly using this horrible turn of phrase of sort of return on AI. Does it matter what the return on that investment is if companies just have to stay in the game no matter what? So I think some companies are really benefiting massively. Right? Like, for example, I mean, they've added $1,000,000,000 in market cap and started growing 20% because of AI. But I think actually for now, the question of AI is too early, and I'll tell you why. First of all, we're going from this world where we have clusters, A.I. clusters of less than 100,000 going to a million. That means we're going to train way bigger models, way better models in the next one or two years. So we'll have to see what these models can do to really assess our ROI. So that's a big question.

The second question is actually if you go back in time and look, for example, at the introduction of Microsoft Excel. Every company in the world had to use it to stay competitive. I'm not sure every company in the world generated more revenue because of it is probably going to be similar. So some companies will generate massive revenue growth like matter. Some companies that are going to be you stay competitive and within within your industry and you have to pass on the benefits to your customers. But you got to do it no matter what. And maybe it comes back to sort of your point, Is this a little bit like the space race 2.0? You got to be in it to survive. Daniel, the story so far has been about the hyperscalers plus meter spending. That's been all of it. And I wonder it's what standard do you hold? Jensen Huang To demonstrate that there's more than that outside of basically just five companies as customers?

Yes. So, I think there's actually way more than that. So, there's some smaller players, so-called GPU cloud companies like Warby, Lambda Labs, their significant customers. And actually, the second thing for some of those Hyperscalers, the use case is not internal, but it's actually external. So, they're buying these GPUs and they're renting them out to companies that want to train some sort of new application, usually sort of venture-backed type of startups. Right. And I think the key thing to look here is when do the Fortune 2000 venture, the big companies start going to these clouds and start training their own models? Now, Nvidia has a solution for enterprise A.I. They're trying to make it easy for them, but we have to start seeing that. Well, Daniel, let me jump in. I mean, I'm trying to actually define what Nvidia is. Is it a chip company?

Is it a systems vendor? And then they have in video CPU cloud. And I look at that and go, Does Jensen Huang just want to be a cloud provider in his own right? Yes. So I think Nvidia is the world's leading system optimization company and it starts with a chip, it continues with the software and now they're going rack scale and cost their scale. And some of those systems have 600,000 parts. I mean, that's a big, big problem to solve. Now, I think what the company is doing, I think the idea is more to be the cloud within the cloud rather than a cloud yourself. So it's actually much more lucrative to sell to many, many parties and get a big fee out of that rather than creating your own cloud. So I do not think that Nvidia will create their own cloud. It's actually just fine to sell through third parties like the Hyperscalers and the cloud, etc., etc.

Daniel Sams Capital has $46 billion in assets under management. There has been so much zeitgeisty-like euphoria around this one single name that many have launched ETFs that double leveraged the particular moves of Nvidia, and you just see the demand. They record shares outstanding for granite shares to X long Nvidia daily ETF. These are exactly what a chart hat tip to Dan Curtis. We thank him. This is all about ultimately wanting to get long in whatever way you can. What are you seeing, Daniel, in terms of your exposure? How long do you get Nvidia? Yeah. So, I mean, well, you know, we'll go over time. Maybe we'll get a bit bigger, bigger or smaller over time. But ultimately, I cannot imagine not being invested in Nvidia over the next 5 to 10 years. And I'll tell you, super simple why. It's pretty obvious that we're going to

need more computing in the future, more computing within health care, drug discovery, self-driving vehicles, factory automation, enterprise, you name it, it's going to go up a lot. And the second thing, it's fairly obvious too, is that in video, as a leading company in the world, that provides compute, right? And there's a bunch of other companies doing it too, but in video as a leader. So if you add those two together, it's very likely that the video is going to be a much bigger company in the next 5 to 10 years. Now, that doesn't mean that we don't have a down cycle at some point. It doesn't mean that we have, you know, maybe a bit of an overbuild at some point. And then we go back to, you know, acceleration. But that's sort of how I think about the longer term. It's the beneficiary of the growth and compute needs in the world.

You acknowledge there are other accelerators out there? Am I 300 X and Gowdy Intel? To your mind, how is it that NVIDIA maintains its position as market leader? Yes. So I think it's a little bit like that Wayne Gretzky saying of go with the puppies. So Jensen tends to be there already and the others are following a little bit. And it's just beautiful. Right? Jensen has seen this 15 years ago. So he started off with a chip. He built this. Don't give me that. It's beautiful. We're talking about computer hardware. Come on, Daniel. You think the same, Ed? You think exactly the same pot calling the kettle black. Keep going, Daniel. Sorry. So it's sort of this idea of, like, you went from one chip to a big system and you optimize that over time. But in order to do that, you need many parts, right?

You need the software, you need the hardware, you need the GPU, you know, they have a CPU, a networking chip, etc. etc., etc. And, and Jensen has seen this earlier than anybody else and that continues to be the way. Right? The others are catching up. They're doing M&A to do that and it's fine. Right? They're great companies. But ultimately if you want the cheapest compute, the cheapest dollar per compute, NVIDIA is the best company to go to and it's unlikely to change in the future. I can give you one example. I think imagine today we got a chip out this table as good as Nvidia and they would ship it to all of us. The problem is we wouldn't know how to use it because the software isn't there to use it, whereas within video they've created entire software for each industry in the world and it's actually easy to use because of that. And we would have to hire people if we

wanted to do a third-party chip rather than Nvidia. And that's sort of the secrets of.

Investors Await Nvidia Earnings After Market Close

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

The main piece of information sought is confirmation of ongoing scaling law improvements in AI, which is crucial for NVIDIA's long-term growth.

The focus is on the increasing compute intensity for training AI models and the evolution of AI within the scaling laws.

The discussion also highlights the uncertain return on AI investments and the transition from smaller A.

clusters to larger ones, aiming to train better models in the next one to two years.

The Impact of Microsoft Excel and Nvidia's Role in AI

Microsoft Excel's introduction was crucial for companies to stay competitive, with varying revenue growth outcomes.

Nvidia's impact extends beyond hyperscalers, with smaller GPU cloud companies and external rental services for startups.

The key focus is on when Fortune 2000 companies will start training their own models in the cloud, as Nvidia aims to make enterprise AI solutions more accessible.

Nvidia's Cloud Strategy and Market Impact

Nvidia is positioning itself as a leading system optimization company, focusing on chip production, software, and now rack-scale systems.

The company's strategy is to sell its services to multiple parties, rather than creating its own cloud, which is more lucrative.

There has been significant demand for Nvidia's stock, leading to the launch of double-leveraged ETFs and a strong long-term investment interest.

NVIDIA's Future Growth and Market Leadership

NVIDIA's computing power is expected to become increasingly important in various sectors, including healthcare, drug discovery, self-driving vehicles, and factory automation.

The company's leadership in providing compute resources positions it for significant growth in the next 5 to 10 years.

NVIDIA's successful trajectory is attributed to its early vision and strategic optimization of its chip systems over time.

NVIDIA's success in computing is driven by its comprehensive approach, offering the necessary software, hardware, and GPU.

The company's early foresight and industry-specific software solutions have created a unique advantage, making it the go-to provider for the cheapest compute.

While others are trying to catch up through mergers and acquisitions, NVIDIA's dominance is unlikely to change in the future.