Jensen Huang’s ‘Digital Twin’, Future of Creators, OpenAI’s International Issue | Jan 5, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Pasricha. It is Monday, January 5th. Happy New Year to you. Hope you had a great holiday. We have got a great show to kick off the new year. First up, we are talking CES. The Consumer Electronics Show opens today in Las Vegas. We'll also discuss what NVIDIA CEO Jensen Huang is likely to focus on. We'll then dig into AI adoption with an expert who helps companies navigate which AI tools to buy and how much they should actually pay for them. Next up, we've got friend of the show and Slow Ventures general partner Sam Lesson joining us to share his predictions for 2026.

0:53And we will wrap with a look at ChatGPT's user base and where OpenAI needs to expand as it looks to achieve its ambitious revenue goals. It is a big show, so let's get right on into things. CES, the Consumer Electronics Show, kicks off today, and NVIDIA's Jensen Huang will be there. One of his big focuses will be showing how NVIDIA's chips and software can be used in manufacturing and robotics. My colleague Wayne Ma published an inside look at that product line and why Jensen has been frustrated with its growth lately. I want to bring on Wayne to talk all about his reporting. Wayne, welcome back to the show.

1:30Happy New Year. Thanks for having me, Akash. Happy New Year to you as well. So, CES. I've never been to CES. You've been a couple times? Yeah, I was there last year. Okay. Okay. And for those of us who haven't been, I mean, you know, tell us a little bit about the vibe. What are people talking about this year? Yeah, I mean, it's like kind of like the annual pilgrimage of consumer electronics companies. But, you know, the last few years, it's evolved into like more than just, you know, gadgets, you know, like your phone, but, you know, automated cars, you know, robotics, AR, VR, and now, of course, AI.

2:07And so the products here, I mean, you know, I always seem to think about what comes out of CES. These products, I mean, like in some cases, they're still like years away, right? I mean, it's never the stuff that is actually scaling now. Or I mean, what do they end up revealing? It's really mixed, right? Like some of it can be vaporware, you know, people announcing gadgets that they claim will come out in a year or two this time, but they never come out. Other things are, you know, coming out soon. So I guess it's kind of like a mixed bag. And from all the years you've gone, do you remember a particular memory from CES that comes to mind?

2:43It's like, oh, my God, that was so CES. Oh, I mean, I think the parties, you know, there's all these like lavish parties, like Instagram hosts a party, TikTok hosts a party. And basically, it's like very, there's like lots of marketing, basically, and companies trying to get on your good side. You know, TikTok definitely trying to push that they're not like a Chinese company, for example. So there's just lots of, I remember a lot of the parties. Right. Okay. Well, okay, let's talk about what's happening on the ground this year. I mean, you had this story out today about NVIDIA's push to get companies who are focused on manufacturing to adopt not just their chips, but also their software.

3:22Talk a little bit about this business line and why this is such a big focus for NVIDIA, and then we'll talk about how it's going in a second. Sure. I mean, obviously, NVIDIA makes billions of dollars selling chips for training and also chips for inference. So you could think of it as like a computer for a robot. You know, a robot needs a computer to train its mind. People are buying chips for that. A robot needs a computer to run in itself. NVIDIA sells the chips for that as well. But to simulate the robot in an environment and collect data on that, to train the robot's movements and things like that requires another computer.

3:58And people don't seem to be buying that type of computer or buying NVIDIA chips for that. and Jensen Wallen has long believed that that's like the third pillar of their business, where they could also make, you know, billions of dollars, but it's just not happening. And so you talked about this idea of a digital twin in the story. So help me understand what exactly this simulation looks like and what it's emulating. Yeah, so a digital twin is kind of like, say you were simulating a robot. You'd build the robot in the digital world. You'd build the digital world itself, and then you would like run the robot in that world and collect data on it to kind of help improve like how it moves and how it thinks.

4:35And so it's essentially like a digital version of the physical object that you are trying to make or build. So it's like putting like a digital version of the robot or the machine in sort of like a metaverse-like reality to emulate what could happen, essentially. Right, right. And so NVIDIA has been pushing tools to not just build these like 3D objects, but also to simulate them. And Jensen believes that everyone in the world will need the digital twin, including yourself. You'll need your own. And of course, you'll need to buy an NVIDIA chip to run that digital twin. So everyone will have their own GPU, their own digital twin.

5:13That's like his vision for the future. That's his long-term vision. So he's been working toward that. And so they announced Omniverse, this kind of division or product in like 2021. And you can download the software for your computer and run it on NVIDIA chip on your computer, but you can also lease or buy and rent thousands of MVA GPUs in the cloud. And so my story basically kind of talks about how people aren't doing the latter, right? They're not using large-scale simulation in the cloud. Why not? I guess there's a lot of reasons. One is internally there's issues. The company itself has had trouble executing, building software for that division.

5:53They've built lots of flashy demos over the years that they've shown to Jensen to try to impress him. but starting around late 2023, early 2024, Jensen was starting to get frustrated because he saw these great demos, but none of them actually turned into real products. And he felt that engineering time was being wasted on these demos that they had to like, they should focus at least and pick at least one thing to actually ship and deliver. And they weren't doing that at that time. Is there actually a market here for these products and tools or is this really just NVIDIA looking for more ways to grow?

6:27do you think? Well, Jensen believes, especially with the explosion demand for AI, that there definitely is a market that eventually, if you're going to have lots of robots walking around or autonomous cars driving everywhere, that obviously you need to kind of simulate what they do. You can't just put them on the physical world. So that's the other way, right? You can just run them in the physical world and collect data for that. But that's expensive, right? You have to build the product first. You have to run it in the real world, especially with autonomous driving, cars could hit people, things like that.

6:57So if you just run them in simulations thousands of times, millions of times, and collect the data that way, it's safer and cheaper. But who are the customers that have bought into this idea of using these simulations so far? Well, lots of people do use simulations. It's just they don't use NVIDIA's products. So Tesla's simulating their cars on the road and Optimus, their robots, their humanoid robot as well. But they don't use NVIDIA's tools. right? They may not use NVIDIA's chips. So, and so in some cases, it's actually that the competitors' products actually work better for this stuff. Yeah, or they build their own in-house because they don't feel like NVIDIA's solution is, you know, right for them, right?

7:39So it's like very different than training and inference where everybody uses NVIDIA's chips, NVIDIA's like the de facto standard for that, and nobody wants to use anyone else's stuff, right? And so I think that's the big struggle for this division and for this business. And so as our reporter covering NVIDIA, you talked to everybody in this ecosystem. I mean, looking ahead a year or two out, do you see this being a major business line for NVIDIA? Or do you think it's kind of just puttering along still? I think one or two years out might still be too early. Maybe five years out, 10 years out, for sure.

8:12But again, the marketing has to mature. We don't have humanoid robots in our homes today like we do with a smartphone in everybody's pocket. And so I think it has to get to that point where everybody's buying a humanoid robot or some robot for their home. And then, of course, this market for simulation and 3D design tools will take off. Right. All right. Well, there's a lot to look forward to in there. So I want to thank you for coming on. Wayne, that is Wayne Ma, our NVIDIA reporter here at The Information. Thanks for having me. Okay. Our next guest has a ground view of how AI adoption at enterprises is looking on the ground.

8:51Adam Mansfield is practice lead at Upper Edge, which helps businesses negotiate big contracts with enterprise software companies like Salesforce and ServiceNow. I want to bring on Adam to help us understand the current state of play of software heading into the new year. Adam, welcome back to the show. Happy New Year. Happy New Year. It's good to be here, kosh always a pleasure i was told by my team that this is the only day that i can say happy new year uh given that it's january 5th and then i'm not allowed to say it afterwards so i'm milking it for for for everything that it's worth all right hey i get it i get it i have similar restrictions at my company so i understand okay so look let's talk about um enterprise software there we we had a crazy year in 2025 we're going into the new year i want to understand a little bit from you first you know, the three of the main companies that you help your clients negotiate deals with are Microsoft, ServiceNow, and Salesforce.

9:47And we've done a lot of reporting on all three of these companies. I want to understand from you how these three companies pitch their AI services differently in the market and what that kind of tells us about just how overlapping some things are in some cases or just how wild west it is in terms of trying to get people to actually buy these products. Yeah, it's interesting, right? So there's a couple of ways I would look at this. Number one, you have to think about these vendors, the vendors you just spoke to about Microsoft Salesforce and ServiceNow, right? So we start with Microsoft. Microsoft's foundational approach to the market is always the same.

10:30We are Microsoft. We already have a base of customers. So they come in with a one Microsoft approach, right? So we're, we're investing, we're AI leaders, you should naturally come and follow us down this path. That's kind of the mindset. The problem is, and there's a lot of reporting out here on this, not just from my own insights, you know, they haven't been able to scale that as much. Okay. So what they've started to do, and this is interesting, they started to kind of transition a little bit away from kind of a headcount base. So Copilot, which is an add-on to your Microsoft 365 subscription.

11:04Copilot, $30 per user per month. It's their list price. They're having a little bit of stalling on the acceleration. Now though, there's other consumption-based products that they've been pushing. Okay. So things like Copilot Studio, that's not headcount based. That is how to build agents outside of Microsoft 365 that essentially is based upon a number of messages. So$200 per month for 25 ,000 copilot. And that's how this runs. So they started to create that flywheel effect that you typically see in AI vendors. The flywheel effect becomes really important because that is how the market's looking at things to accelerate revenue growth tied to AI.

11:49So Microsoft is starting to transition there with that. Okay. I would also say, and this is a really big one, Microsoft 365 E5, which is a foundational suite of products that they've also been trying to push onto the market. They haven't really gotten where they want it to go. They have now included something called Security Copilot as part of E5. Now, part of that's to justify the July 1st price increases that are coming. But the thing that's really, really interesting here is somewhat of a hybrid. So you still have the Microsoft 365 E5 SKU, which is headcount based. But within it, you now get a number of security co-pilot, which is the security.

12:29So you get more in there, essentially. Exactly. But here's the hook, though, Akash. It's really important. Those SEUs, you only get so many. And if you don't, if you blow through them, guess what? You owe them more money. So they're now in a hybrid, which becomes very problematic for customers. But this consumption-based pricing and usage-based pricing that you talk about here, I mean, this is uniform throughout for ServiceNow and maybe in Salesforce, right? Exactly. So that's where I'm going. It's exactly. So Microsoft led with, for the most part, led with Copilot, which is an add-on to the headcount base.

13:05And now they're starting to mix it a little bit. Salesforce, right out of gate, consumption-based, flywheel-based, right? Which is AgentForce, as well as DataCloud, which they always push together as one thing, okay? So agent force originally was$2 per conversation. The market didn't like that. Well, it's a conversation. I don't understand. So then they've transitioned to flex credits. Well, flex credits, you get a certain amount of credits driven by a number of actions. Again, what's an action? And I can go down that path, which I'm not going to do now. But that's all consumption-based, all consumption-based.

13:40And that is really being driven when you talk about uniformity. Yes, it's to drive revenue trying to do this in the market. Investors like that. But the other reason why you have to think about it is that these AI, these cloud vendors, they have accelerated cost profiles driven by the fact that their cost and structures with those underlying foundations are scaling based on real-time usage as well. So to maintain their margins, they have to sell products that are consumption-based because that's the pain that they're feeling to their cost profiles. So let me ask you a question about then consumption-based pricing and the impact that it has structurally on enterprise software.

14:21And you've been on the show before, and we've talked about this idea that, well, if the number of seats go down, how do you make up for it? And consumption-based pricing might be the answer to that. But one of the questions I have for you, Adam, is with seat-based pricing, I mean, there was this whole idea that, hey, I just want to pay for the one vendor and I just want the one seat. And I'm not interested in going to all these different vendors. Just give me an all in one sort of platform. And now in a world where it's, well, I'm only paying for what I use. I'm sort of torn on whether or not there will be a fragmentation of of services because people won't really need this all in one platform.

15:06or are all these products overlapping and what's the point of fragmentation when there's going to be consolidation anyway? Yeah, so there's a lot here. I was told I only have so much time, so I'm going to keep myself honest in this. No, no, no, no. Here's the answer in its simplest form. I believe it's still going to be very much a mix and a problematic mix for customers. And where I'm going with this is you're going to have some AI products that are going to be still headcount or hybrid, like ServiceNow, you mentioned before, ProPlus, that add-on to get the number of assists. You still have a unit, and then there's a number of assists, very much like I just talked about with E5.

15:48The big thing that's going to happen is these organizations are going to have to create, and when I say organizations, the customers, they're going to have to create governance models just literally on how we are tracking our actual usage at the seat level, like Bob in accounting or Sally in marketing, they need to be accountable for what they're using. Because think about it, when the consumption happens, it's not just one thing that Sally's using. Sally has 15 different applications, all different types of meters, and they have to all be organized and then sit back at a corporate level and govern because those invoices are going to come from those vendors, whether it's Microsoft, Salesforce, ServiceNow, SAP, Oracle, all right.

16:34So I don't think it's going to create an environment where, oh, it's so much easier now. I actually think you're going to have to have dashboards. You're going to have to have governance, governance teams, quarterly meetings. There's going to be a whole host of things that are going to have to be put in place, trying to track what is actually being used. And does it? I mean, from what you're seeing from your customers, is the product working? I mean, all right. So here's what I would tell you. I would tell you that I think there's some momentum to, okay, I kind of see where this is going, seeing where it's going.

17:08Here's the problem though. Seeing where something's going when you're cutting checks for millions of dollars for the thing that you're still wondering where it's going is a problem. The other thing I would tell you is that, is it working? Well, the first thing is we need to step back. How is it actually being used? This is the problem. The definitions of use and you go back and say, well, are you using it? Someone doesn't even know what use is. Well, I don't know. I had six conversations. I had five actions. I had, I don't understand what that is. So now you have this mix. So the fundamental question, is it being used?

17:43I would say it's being used not to the level that it was expected, number one, but number two, I don't think internally they've even figured out what use actually is, let alone ROI, productivity, all the things that the C-suite's expecting. And lastly, the problem is these vendors are still stuck in the mindset of push, push, push. Okay. I want X volume. You need Y number of actions. And they sell to this outer level to drive revenue. Here's the problem. When you challenge that vendor to say, how did you come up with that number? How did you come up with number of actions that you think I need?

18:21How did you come up with the number of credits? Dead faces. Maybe that's a bleak term to use, but the point is there's no way for them to actually define it because they're figuring it out too. So really, do I see use? I see testing. I see figuring out. I see willingness to explore. Am I sitting here with customers like, wow, I got this figured out. I love it. I see the ROI, let's do more. That's far and few between. So let me ask you one more question before you go. So now with the customers you're talking with and representing in these deals, what do they want more of from these companies? I mean, the products are there and they're being packaged every which way and priced a number of different ways.

19:06And, you know, like what do they want more of? Yeah. All right. So the first thing, honesty. They want them to come in and say, look, we ourselves are figuring this out. We ourselves are figuring out what the actual specific ROI productivity is going to be for your defined use cases for you. Lead with that. Honesty, work together, workshops, spend your time focused there. That's what customers want. Get over that hurdle, do the demos, see the value and say, okay, now let's talk honestly. Let's commit to this amount with full transparency, which is the second thing they want. I need full transparency.

19:55What's used? What's an action? What's a conversation? What's an assist? I want it all defined for my scenario as you understand it today, not generic PDF PowerPoints that you send to everybody or I can go to your website. I don't want that. The third thing, cost certainty. That's obvious, but let me play this down for an extra tick. One, it's upfront. But two, I want you to tell me what does the world look like when the flywheel actually spins? The thing you want, which I may want too, what is it going to cost me? What assurances, what volume structures, what credit structures, what type of commitments, protections are you going to give me?

20:34Fourth, last thing I just thought of, investments. They need investments. These vendors need to come in and say, look, I know it's going to be change management. I know it's going to take time. Let us invest. Let us give you resources. Let us give you funding as investments to make the thing real working together. Okay, and do you think we get that in 2026? Yes, no, maybe. I think we're going to get closer to it. Do I think we're going to get all of that? No. Do I think that vendors are still going to hold the line because they're stuck in where Microsoft, where Salesforce, you should follow me where I go?

21:12I do. Here's the problem. In 2026, I do think there's going to be vendors that are going to come around, whether it's a Sierra for Asian force competitors or others that are going to come in and they're going to start accelerating their own usage, which then gets them in creating their own moats. So do I think? Yes. I think there'll be more. I think you're going to see better behavior. Do I think they're going to solve it willingly out of the gate? No. And I think you have to know how to orchestrate those conversations way in advance, which also not, you know, that's something that we do. But the point is customers need to do this with or without the help.

21:46They need to make this their foundational approach to get behavior change. They have to force the behavior change, the customers, because the vendors are not going to do it. Because remember, these vendors we're talking about, they report to one group, the financial community, their investors. That's it. They can talk about customers and how much they care about the customers, which I'm not saying they don't. But at the end of the day, they're publicly traded. Let's just be honest. They need to drive revenue, period. Right. Well, Adam, I want to thank you for coming on. I appreciate the time. That is Adam Mansfield, a practice leader at Upper Edge here on TITV.

22:23Always a pleasure. Happy New Year. Happy New Year. Okay. Our next guest is a friend of the show. Sam Lesson is a general partner at Slow Ventures. He is also the husband of Jessica Lesson, our editor-in-chief. Sam put out a pretty interesting set of predictions over the weekend over how he thinks 2026 is going to play out. It covered everything from degen culture to wealth creation to relationships and AI. I want to bring him on to talk all about it. Sam, welcome back to the show. Always happy to be here. Happy New Year. Happy New Year. Okay, so let's talk about this note you put on. Look, you talked about default depression.

23:01You talked about the challenges of wealth creation. You talked about gambling. It felt like 2026 is looking a little bleak. I don't know. Hey, you know, I was flying home for vacation and I was trying to think about what this year is going to be like and what really matters thematically in business and culture and society. And look, I think here's the interesting thing. We live in a world ruled by narratives, especially now, especially given how they travel on the Internet. And there's a lot of things going on which may or may not actually fully be true yet or really ever. But people believe them.

23:31And the fact that they believe them then drives their behavior. right and so i think you know a really good example of that is this narrative that ai is going to replace all the white collar jobs right um now there's is there some shades of truth to it probably is it this year probably not and yet it's going to color how people feel about their lives how they behave what they do and the question you know from my perspective what i was kind of exploring is like look you know given the narratives that are happening right now in the world, how's this year going to feel? Okay. So, and I'm curious, by the way, you put these notes out.

24:06Did you get any responses to this one? I mean, what do people say to you when you put out these notes? I got many, many responses to it. It's through my newsletter. And, you know, it's one of those things where you only hear from the people that agree with you. So the people that agree with me, agree with me. And they're saying, this seems pretty depressing. And, you know, how are you going to coach your kids, et cetera, et cetera. I don't have any great answers to that. But yeah, there are people, people definitely, I think the real, the biggest big picture here is, again, thinking about this year as a year, which even aside from technical realities will be ruled by how people, by feelings.

24:37We're in a feelings culture. We're in a feelings year. There's a lot that is, from a narrative perspective, could really end in lots of different places. Like, we don't know. We're kind of on a knife's edge in terms of how the year plays out. And so I think the sense is everyone kind of is thinking about how to react, you know, in real time versus thinking long term about what's actually going to happen. Now, what about prediction markets? I mean, I assume that's what you were pointing to with your note here about gambling. And I think you were basically drawing the connection that, hey, it's really tough for young people to make money nowadays.

25:07You know, maybe that has something to do with all of the momentum and prediction markets. Is that the point that you were making? Well, it's complicated, right? Like, look, I actually love prediction markets. I think prediction markets are awesome. You know, I'm like a student of Hayek, right? And markets are information machines. And especially as traditional sources of news and trust and truth fall apart, the idea that the best predictor you have of what's actually going on in the world is going to be through a prediction market and mediated by a market. It's a great intellectual theory. It's great.