The Six Five team discusses NVIDIA GTC.
If you are interested in watching the full episode you can check it out here.
Disclaimer: The Six Five Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.
Patrick Moorhead: Gosh, where do we start? So there were 36 press releases and blogs and GTC happened over four days. I’m going to hit three topics and then put some macro ideas on there. I had a good talk with CNBC street signs on Tuesday night, but I’m going to bring it all to you. So some of the highlights that they brought out, first of all was NVIDIA DGX Cloud. And this is essentially customers getting access to DGX capability through Azure, Google Cloud, and Oracle Cloud. NVIDIA actually recognizes the revenue and it runs NVIDIA’s enterprise software stack. This is a huge one for a lot of reasons at primarily because this is not, let’s say Azure on DGX Cloud. This is DGX Cloud on Azure and I think that’s a big deal because it’s being sold by NVIDIAns.
Apparently NVIDIA, CEO Jensen Wong did say, “Hey, of course the Cloud Salesforce can – Cloud meaning Azure, GCP and OCI – can sell this, but we’re going to be recognizing the full revenue of it.” Call that account control, call that whatever you want. This is a step in vertical integration of the company. The other element of vertical integration is what is called NVIDIA AI foundations. And this is a PaaS play out there. Again, the first one is an IAS play. The second one is a PaaS play, which is offering their services and this one for LLMs, visual language models, and models for drug discovery. Each of these have fancy names. Look them up in the show notes if you want to know-
Daniel Newman: One of them called Picasso.
Patrick Moorhead: I think it is. And NeMo was another one. I think that’s the one for drug discovery. But essentially these two together signified to me a vertical integration play. And on a call with industry analysts, Jensen, he really laid it out there. He talked about what the company did in gaming, which was they focused on the overall total experience and then that experience was distributed through a third party.
This is what he is bringing to the cloud. His timing is perfect. Quite frankly at scale, there is not another training solution out there. That’s just a fact. Not saying that the companies like Intel and Groq and Tenstorrent and folks like that don’t have a chance, but that’s what’s installed right now. And on the inference side there’s a lot more competition out there as well. But for the sake of enterprise and at scale, it’s still an NVIDIA game. The only caveat I’ll say is that statistically most inference is done on a CPU, which benefits Intel.
So with all that caveating said, I think this is a big move for them. You can only make a move like this when you are at the top of your game. And right now NVIDIA, whether it’s hardware but more specifically CUDA software and the de facto standard it has become, this is where you can pull this and I think sometimes you create your own luck. Sometimes timing is there and we can all debate whether NVIDIA brought generative AIs to the table or that was recent – doesn’t matter. They’re at an optimal time to monetarily, financially take advantage of this. Not just slinging hardware, but now bringing IaaS and past services to the table sold by their direct Salesforce.
Daniel Newman: You know what? I think there are a number of instances in this particular space where there’s going to be credit attributed, meaning who founded generative AI models. NVIDIA has certainly been pioneering, it’s been a bit of a one track mind in its existence, whether that’s using GPUs for gaming and generating ray tracing capabilities and next generation AI to make gaming experiences better or data center AI. But the theme has not been wrong. And Pat, there’s nothing I like more on the Six Five than doing a victory lap. And so I just want to say that in August of 2022 and NVIDIA stock had fallen to 130. By the way, it’s twice that now. I went on CNBC on Squawk Box and I said, NVIDIA’s going to have a huge year because of AI. AI is going to be massive. This was before GPT was being discussed publicly.
This was before generative was even in the minds of most people. But the trend line was obvious. And the reason I point this out is there are over rotation a little bit. The fact that there it’s a joke at times. The market cap of the company and the multiple that it gets versus say an Intel, you’re talking about 10 versus hundreds. But it is the perfect identification of when a company is attached to the right trend lines, how its value can be perceived. This GTC was all about AI. It was all about making sure whether you can consume it in the enterprise, in the cloud, it’s attaching to the hyperscaler, it’s attaching to data center. It’s making sure that the entire world knows that it is a combination of capabilities that NVIDIA has been steadfast to build over the last several years.
Hardware, software, frameworks and ecosystem is going to be paramount to the development of a vibrant AI capability. And all this generative stuff we hear on largely does run on NVIDIA. It’s like 90%. It’s a huge percentage of what we’re seeing right now. Having said that, I do think there’s some vulnerabilities that are going to be exposed over the next few years. I think first and foremost is cost is going to become a heavy sensitivity and NVIDIA is the premium product. Now Pat, you and I have no issue with being the premium product. I think you are the uber premium product, but in all seriousness, I appreciate the companies that want to make the best and charge appropriately for it.
But there is an entire ecosystem of startup companies and competitors in the chip space that are going to be coming after NVIDIA. And if they can build software like the Groq conversation we had last week where you can compile and move a large language model onto a different piece of hardware without a lot of resistance, you will start to see companies, even the Microsofts and Googles going, “How much can we spend to actually deliver all of this generative AI to our customers?”
So that’s going to be a really interesting inflection point. The second thing is of course I think this the entire GPU and its impact on power consumption and water is going to be really interesting, Pat. You and I have been following, had some interesting recent conversations. One with a company that’s actually making it rain. There is a water issue and data centers are a massive hog of power and GPUs tend to be not the most efficient way to deliver AI. So Asics are going to become really interesting over the next few years. So it’s going to be kind of interesting to watch ’cause you’re not hearing much from NVIDIA about where they intend to participate in any application specific. And then maybe the final thing, Pat, and this is a little bit more of a macro theme than I’m talking about, but that GTC didn’t cover and that’s the skilling issue.
Have you ever seen a technology in your lifetime that’s been as rapidly onset and potentially disruptive to jobs as the last couple of months of generative AI? Now again, this is not necessarily specific to NVIDIA, but to all the companies including NVIDIA that are building the technology that are going to rapidly onset and displace certain categories of workers – what is the responsibility of the industry to help that particular class of worker to find their next thing? And this is something I’ve talked to executives at Microsoft about, at Google about, AWS about, and NVIDIA I think has a role to play because frankly NVIDIA’s great innovation has been a catalyst of the speed of all these next generation AI models that we are experiencing.
So Pat, I’m going to leave it there. Another big week. NVIDIA, we got a huge bump, huge run up in the market and that’s because there’s nothing hotter Pat, than AI right now. And what did I call it? The gift that keeps giving our show is going to be the week after week after week of discussions on generative AI. And guess what I don’t expect is unlike the metaverse, which by the way NVIDIA is also enabling the actual usable digital twin metaverse, I expect this trend is going to have a staying power that is different than crypto and metaverse in some of those of the past.
Patrick Moorhead: Yeah, I’d like to talk more in the future about reskilling. Back in the eighties and nineties with some of our international trading policy, we shipped a lot of manufacturing jobs overseas to get prime terms. And neither those companies, the Ford Motor Companies, the GMs of the world, the steel companies, did any of the reskilling and the government didn’t come in and help either. While I rarely trust governments to do the helping hand efficiently, this could have been an area that could have done that. So if we believe that AI will replace jobs as opposed to – let’s say market expansion – people will need to find other things to do. That’s just kind of a true statement. So let’s keep that conversation going Daniel.
Daniel Newman: Yeah. Absolutely Pat, and I’m glad that you mentioned that because one of the things for me and for you and I and just being in this space is that we do have a certain amount of responsibility as we are the arbiters of good and bad in tech is thinking about the unintended consequences, because there’s a lot of great intended consequences. Look, I couldn’t be more excited about some of the things generative is going to enable me to do in my day in and day at work and enable for our businesses to do. But I also think about what about all those agencies that we hire or those contractors that we use to do things that we could potentially not require anymore. What happens to them? If I wrote a book, by the way, it’s been translated to Chinese now – this is Human Machine – about this.
Patrick Moorhead: That was the best pat on the back I have seen. That’s so good. I love victory laps, don’t get me wrong. But you have the physical, tangible – it’s like these ribbons behind me if I actually won anything behind me.
Daniel Newman is the Chief Analyst of Futurum Research and the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise. Read Full Bio