On this episode of The Six Five – In the Booth, hosts Daniel Newman and Patrick Moorhead welcome Jonathan Seckler, Senior Director, Product Marketing for Dell Technologies.
Their discussion covers:
- Dell and NIVIDIA’s partnership announcements at GTC this week
- How Dell and NVIDIA, in collaboration, can solve key critical issues with AI
- The applications and use cases that will benefit from AI and GPUs
- How customers can gain an edge in the race to support generative AI
Be sure to subscribe to The Six Five Webcast so you never miss an episode.
You can watch the full video here:
You can listen to the conversation here:
Disclaimer: The Six Five In the Booth 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: Hi. This is Pat Moorhead. We are back for another Six Five Virtual In the Booth recording with Dell, talking about NVIDIA GTC. Daniel, here we are again. GPUs, generative AI, my gosh, can we have any more of the coolest topics out there? I think three quarters of our Six Five topics, the last two weeks, have been literally all about this.
Daniel Newman: Yeah. It’s really hit that inflection point, Pat, and it’s going big. A year ago, we were looking at the market, trying to figure out what’s going to be the big wave. We’ve had the Metaverse and many of these other things. This one’s moving, but not only is it moving, but it’s really practical. We’re not getting just announcements. We’re getting announcements, and we’re seeing things, and it’s been really eye-opening, Pat, and it’s really exciting. GTC is just a big moment, a big opportunity to get this story going even faster.
Patrick Moorhead: Yeah. Dell has always been a player in this space, whether you would want to call it a high-performance computing or AI. It’s my pleasure to introduce Jonathan Seckler, who is a third-time Six Five guest. Great to have you back.
Jonathan Seckler: Yeah. Getting up there. Thanks a lot. I commit to you, today, that everything we talk about is not coming from ChatGPT, so this will be real content in this one-
Patrick Moorhead: Good. That’s great. Thank you.
Jonathan Seckler: That’s what ChatGPT would tell you, though.
Patrick Moorhead: Yeah. Well, I’m actually being rendered by a generative AI natural image video, so it’s really not me. I’m getting a tan somewhere else.
Daniel Newman: Pat, we’ve done a thousand of these. Don’t you think that one of these large language models could process all our video, and then basically start to generate new videos based upon our catchphrases, the way we opine about certain topics, connecting the back end to the front end? Things that we say a lot. I think, eventually, this will be the deep fake. This is a cool way to jump in here, and saying, someday, will it be the Six Five, and it’s not really Pat and Dan, but it’s Pat and Dan and…then you’ll need an AI to discern whether it is Pat and Dan because that’s where we’re going. We’re going to have a lot of fun here. Let’s start talking a little bit about what Dell’s up to. Dell is supporting GTC to talk about its partnership with NVIDIA, the solutions it has for AI. Talk a little bit about some of the key issues with AI that you and Dell are trying to sell.
Jonathan Seckler: Yeah! I think this has definitely been the year for generative AI, and you’ve heard a lot of great stories. I think that there’s a lot of generative AI as showhorse, or generative AI as a play thing, a neat demo. It’s really interesting to see that, but I think, in order to get past that kind of hype-cycle idea, we’ve got to look into what it would take to actually build out and create solutions that can deliver this in an enterprise or a commercial context. At Dell, we look at a lot of things around AI. In our AI solution area, we’ve been talking about validated designs for AI, where we’ve built reference architectures, and we’ve partnered with companies, like NVIDIA and Vmware, on a variety of ways to deploy AI.
I think the challenge here, with generative AI and with AI in general, and what we want to talk about today, is how can we reduce the time-to-model? How can we reduce the time to deploy an AI solution in your enterprise? Those are the big areas that we’re looking at, and I think that’s going to be the challenge, is to make this real, because the challenge or the worst case, here, is that this just becomes another interesting toy that quickly jumps the shark, and on Twitter, and we never hear from it again.
Patrick Moorhead: Yeah. Jonathan, I appreciate you grounding us, too. Quite frankly, just like we saw AI come in, it didn’t mean analytics went away. Right?
Jonathan Seckler: That’s right.
Patrick Moorhead: Very rarely in tech does anything go away. It’s really a matter of “and”. We have analytics. We have certain flavors of AI, and we have this new flavor of AI, but the unspoken thing in the room is, “Well, wait a second. How do we do this software? How do we actually get this heavy benefits, enterprise benefits, and enterprise value?” It’s one thing for, maybe, five companies to be able to do that, but it’s entirely another thing to help democratize this for other enterprises. Are you announcing things at GTC around this simplification, this time to implementation?
Jonathan Seckler: Yeah! We are. A lot of the things to think about, I think, when you think about generative AI is that, right now, all the VC money might be going toward the next big idea, but I’m going to take the opposite tact and say, “Let’s think small.” That’s “small” in the sense of unimportant, or “small” in the sense of, even, low capacity. Think in the terms and the context of your enterprise. The use cases for a generative-AI-type model deployment is more than just writing blogs and faking analyst podcasts. Just think what you can do in a research setting, being able to find the right set of research to build new medical models, to build new technology in a research setting. I was thinking the other day, with all of the news in the world, about a variety of things that governments and law firms can use generative AI to understand the right set of case laws to find the precedent for things that they need to do.
That’s really where we’re focused on, here at Dell Technologies with AI, is how do we help customers build that innovation into their businesses, give them an architecture, give them tools that will allow them to build out those models at the scale that they’re at, and that can grow with them, but doesn’t have to be something where you have to put all of your proprietary or sensitive data out into the cloud where anyone can work with it? Specifically, what we’re announcing here, around GTC, is a number of things. The first thing, the last time we talked was at Supercomputing, I think, Pat. We talked about how Dell has expanded their portfolio of servers for AI. We call it the XE series of servers.
The coolest one that we announced was the 9680. It’s an 8-GPU, high-capacity, NVIDIA H100-based platform. What we’re announcing today is that, now, this product is ready to ship. We’ll be sending out products. It’s available for sale, and is shipping by the time you get home from GTC. It’s a great, I would say, central piece to that large language model deployment that you’re going to need to build for your next generative AI project.
Daniel Newman: It was fun to listen to you say, “think small,” but then you really followed it up by thinking big. Frankly, using the open internet and creating a large language model has been a massive task, but at some point, it actually does become somewhat commoditized. Everybody will get the same answers. We’ve all seen that thinking big is all about the incorporation of your vast data landscape, both structured and unstructured, created in real time, and of course, in your systems of record, and in your storage, and across the web. It’s interesting because I think it’s the “marketecture,” the marketing of generative AI is incredibly impressive, and by the way, very exciting. Look, I never intend to build another PowerPoint presentation again. I never intend to hire someone to build another PowerPoint presentation again. I’m going to hopefully just talk to my machine and be like, “Write a PowerPoint that fits my style of presentation, using this marketing spec,” and off we go! That’s cool!
Having said that, I think some of the things you mentioned, drug discovery, any sort of important quantitative research, it can be used… It takes high-performance computing. It takes AI. It takes the technology that you’re building. Talk a little bit about that. I’d love to understand, in your mind, beyond this kind of hype cycle, which is real, but it is also hype, what is the other applications and use cases that you really see benefiting from AI in GPUs?
Jonathan Seckler: That’s a great idea. When you talk about how applications are developing today, almost everything is using some kind of AI technique to go in. I think as, Pat, you mentioned earlier, it’s like, just because we start talking about AI doesn’t mean analytics went away. Big data still exists. It’s just not that big anymore in the scale of everything we can do. With these solutions that we’re bringing to market, like the 9680, that’s obviously our flagship AI-optimized server, but you can get similar results with any one of our ordinary powered servers or VxRail hyperconverged infrastructure with the right NVIDIA technology embedded in it, with NVIDIA AI enterprise stacked on top of it. You can start to see some of these applications come to life. I see a future where this idea of generative AI becomes really useful in helping marketers, as you said, come up with solid content that will be addressable and usable in the market, but more importantly, that businesses could come up with the kind of code bases, new materials in science, like you said, new drugs as well.
These are all really big ideas that start from some kind of a known basis. The important thing about doing all of this is that there’s always this risk of starting. It’s the fear of starting or the risk of starting that really, I think, holds most businesses back. We’ve seen that to try to deploy an AI solution, a bespoke kind of thing based on your unique situation, can take three to nine months or longer. That’s why we’ve developed these reference architectures. The Dell Validated Design for AI has reference architecture use cases for classic machine learning, deep learning, and just most recently, conversational AI, kind of, generative AI’s little brother of natural language processing, speech-to-text, text-to-speech, and those kind of things.
I think that that’s why I say, “Start small.” We’re not going to be able to create Michelangelo’s Last Supper out of AI on the first day, but you can improve what you’re doing. That is the challenge. Businesses have to innovate. You can’t stop innovating. We’ve spent three years throughout the pandemic, and all the churn that happened afterwards, et cetera, all about transforming IT and transforming business into something that is digital, that is in real time, and that delivers on-demand results. You can’t just stop now, so we need to use these new technologies in order to keep that momentum going.
Patrick Moorhead: Jonathan, it’s one thing to show up on the field, and show up and there’s early majority. There’s everybody on the adoption curve, but how do people get an advantage here, whether advantage over their competition, or be able to change a business model to put the hurt on the competition? How do you get an edge, and how do people get started with you?
Jonathan Seckler: Well, I think the easiest way to start, or… It’s a great question… is that we need to help the customers build their solutions and preferably build it into the environment that they’re already used to. This is why we started our first Validated Design on AI, based on both VMware and NVIDIA. The typical organization out there, most companies, commercial organizations have VMware in their environment. Most commercial organizations, if you’re going to be working in AI, NVIDIA is definitely the leader in this space. By building AI into your existing infrastructure with a few additions to up your GPU game, you get that time-to-model down pretty significantly. You don’t have to start over with a separate environment that’s separately managed, that is isolated from the other data sets in your organization, potentially. Right? You’re in your data center. You have the environment available there, and you can harness the power of all of these use cases that we talked about in much quicker time-to-value than you would if you started from scratch.
Daniel Newman: Well, Jonathan, this has been a lot of fun. I’m very excited to see all the advancements that continue to come out from Dell and its high-performance computing in AI. It takes a village, and Pat and I talk about this a lot. Certain companies, at different times in these cycles, tend to get a lot of credit, but there are a lot of companies that are participating in different ways, whether that’s the picks and axes, the shovels, the plumbing, or that’s those front-end applications that everybody is like, “Oh, my gosh! This is so cool!” Well, yeah, but it might be running the same model that a thousand other applications are running. It’s a lot of different parts, a lot of different pieces, a lot of different community. I think all that is evident every year at GTC. Jonathan, just want to thank you so much for joining us here for the Six Five in the Booth.
Jonathan Seckler: Yeah! It’s been a real pleasure. Like I said, I promise this is real.
Daniel Newman: We absolutely agree, and Pat and I tend to be often arbiters of what is and what isn’t real. That’s our job every single day. I think what’s happening in AI is, and I think you’ve got to continue to follow us here, on the Six Five, to hear more about it, but Jonathan, we’ve got to let you go here. I want to say thanks to everybody for joining us for this Six Five in the Booth. We are talking GTC 2023 NVIDIA in the Dell Partnership. Hit that “Subscribe” button. Join us for all of our other videos at all the big tech events, and of course, our Friday weekly show. At this moment, for this episode, it’s time to say, “Goodbye.” See you all later.