O'Donnell highlights Nvidia's release of its NIMS technology, which he states will provide companies with software that eases the process of building generative AI applications. This development opens up new opportunities for companies to venture into the AI space and further establishes Nvidia as a software provider, unlocking "more revenue sources."
O'Donnell notes that Nvidia's new software releases are designed to empower "regular companies" that may not have the same resources and data scientists as larger companies, enabling them "to be able to pull some of this stuff off."
So clearly, they do have this incredible momentum. And yeah, I do think they delivered on what they promised. And I'll even add one extra thing to what Dan had brought up in terms of important things. Not entirely sure how the market's reacting. Maybe some of this was already baked in.
But look, at the end of the day, this next generation Blackwell architecture sets the stage for them to stay ahead of the competition. And arguably, it was the first one designed and released in the generative AI era. While previous NVIDIA chips, of course, are being used for generative AI, they were actually built prior to the explosion of all this.
This one kind was built in that time era. The other big takeaway-- and this is a big strategic move, long term for NVIDIA like Dan mentioned for Groot-- is software in general. They unveiled something called NIMS-- NVIDIA Inference Microservices.
Very technical thing. But it's these little containerized applets that companies can use to put together gen AI applications. And oh, importantly, of course, it's based on NVIDIA'S CUDA, which is this software layer that's been the sticky part of this whole system.
So the thing about NIMS is that companies can make the process of building Gen AI apps, which by the way, is still really hard, much easier thereby allowing more companies to do this. And oh, by the way, along the way, NVIDIA becomes more of a software company, in addition to being a chip company with new revenue sources from that software.
- And so Bob, how does NVIDIA add on to what is already seeming or feeling like these peak levels of generative AI mentions? Second highest number by the way according to FactSet data of S&P 500 companies citing AI on earnings calls over the past 10 years.
So how much more difficult does it become for, not just NVIDIA, but for the rest of the potential benefactors in this environment to now show Wall Street, to show investors that it can actually deliver on some of these promises?
BOB O'DONNELL: Well, look. Right. That's a great question. And that's why I mentioned this NIM software and another concept they called AI Foundry. And the AI Foundry is this idea of being able to build for people like a foundry, a chip foundry, but this is a software foundry-- these applications because-- yeah.
Look. The excitement's been out there. Now, it's about, hey, we need to deliver this. And we've seen the potential. We've seen proof of concepts. But how do we get regular companies to be able to build this stuff?
They don't have all the data scientists they need to be able to pull some of this stuff off. So how do we make the process easier? And that is what a lot of this software work that NVIDIA is doing, is working on.
And, you know, and on the hardware side, the other thing to remember-- like look, when you're 90% of the market, and you're way ahead of everybody, you're basically competing with yourself. That's what they're doing right now is they're competing with themselves, pushing themselves forward.
And, you know, I think they did a good job with Blackwell. There's a lot of interesting things about the architecture that they've done. Like I said, that are kind of built in this era of Gen AI applications.
And they made some tweaks to allow performance improvements at a big level, and also important energy efficiency improvements, which also is going to help them in the long run.
- Bob, how critical is it for other tech players when you see the reaction in a name like Dell here this morning after Jensen Huang touted the fact that-- he said, quote, "nobody is better at building end to end systems of very large scale for the enterprise than Dell." We're also seeing reaction in SAP this morning. How critical is it for other tech players to be using NVIDIA's platform. At this stage in the AI cycle?
BOB O'DONNELL: Well, you know. I think they all think it's incredibly valuable and important. I mean, I have never seen more partner announcements at an event, I think, in my life. I mean, everybody and their brother like wanted to be associated with NVIDIA. And that's what we saw.
We saw all the major cloud providers, all the server makers, Dell, of course, being called out. But there's also HP, there's Lenovo, there's Supermicro, there's these other companies who are also going to be building stuff with these parts, all these software companies-- Ansys, Cadence, all kinds of big companies who are building software based on this because they see where the market is going.
They see, in fact, that NVIDIA is in the lead so of course they want to associate themselves. And that's why there was just this incredible array, as I said, of different partner announcements because they want to do that.
Now, we're also going to see a lot of competitors, right? AMD is going to get in here. And this market is going to continue to grow. And even if NVIDIA loses share, which is kind of hard not to do when you have that much, the market is going to still grow so much that there's still growth opportunity for NVIDIA.
I don't know why the market's reacting the way it is now. Like I said, hard to predict under that particular detail. But in the long run, I think they've positioned themselves very nicely.
- Hey, Bob. Just lastly while we have here. I mean, there's a lot of focus on what the growth looks like for NVIDIA especially outside of the US as well. Are there any clear answers that are coming forth from this conference about how this company is going to service into China where there's already even more between the US and China of the, kind of, tit for tat that's continuing to play out which puts a bit of a dampener on the opportunities in that region?
BOB O'DONNELL: No. And absolutely, it does, Brad. And I don't see an easy answer. I mean, this is a difficult problem that I think all tech industries are going to, all tech companies are going to face-- people like NVIDIA and AMD and Qualcomm and Intel in particular producing very highly performant chips.
They're the ones who are obviously being restricted more already. And so, yes. That is a bit of a damper on the business. So they are all working on developing chips specifically for the China market that can still provide some capabilities.
China says they want to do all their own, but they are years and years behind the most advanced technologies. So, you know? It's going to be a very interesting situation to watch.
Again, I think longer term, what NVIDIA tried to do at this particular show is show to the world, hey, we're going to build these amazing chips. And we're going to make them better and better. But hey, check out this software stuff because we're going to be doing it too.
We're going to start charging on a per-hour basis, or per GPU per hour to use some of these services. Again, that's really interesting. Revenue there, obviously, almost nothing right now. But what's the opportunity?
- All right. Well, Jensen Huang, CEO of NVIDIA certainly resembling something of the Greatest Showman right now as this event continues to roll on our own Dan Howley out there. Bob, Thank you so much for taking the time as well. Bob O'Donnell, TECHnalysis Research President and chief analyst.