Nvidia’s $2B Nebius Deal, Oracle’s Q3 Comeback, OpenAI to Launch Sora in ChatGPT

20:22Stephanie Palazzolo:And are there caveats to that language that they use?

20:25Tyler Radke:Yeah, I mean, look, I think it's definitely a marketing ploy. Oracle really missed out on the early days of cloud, right? certainly making up a lot of ground here when they're growing their IaaS revenue over 80%.

20:40Tomasz Tunguz:But they are relatively price aggressive on the traditional Oracle Cloud services.

20:46Tyler Radke:Think of like these CPU workloads, storage servers, databases of service. But I do think they have some unique differentiation. I mean, if you think about Oracle's history, they have a lot of hardware DNA, right? They bought Sun Microsystems back in the day. They really understand how to run these engineered systems, parallel workloads, and have a lot of interesting technologies around their interconnects in memory. And so I think as they apply that to the cloud, and again, they sort of designed this new version of Oracle Cloud from a clean sheet of paper. They hired some of the leaders in the industry.

21:26Tyler Radke:One of them is their co-CEO to really understand what were the key learnings from the traditional hyperscalers. So I think based on our work, there is some advantages, certainly, to using Oracle Cloud, both for traditional workloads and for AI workloads, certainly, as you see with some of those contracts. But I will say, like, we do hear they often discount pretty aggressively to get in the door, which makes sense as they're trying to make up for lost time.

22:00Stephanie Palazzolo:Right. And last question for you before you go, you know, there has been this question around the financing chain that exists with their customers. We got some reassurance last night that, hey, because OpenAI got funding, I don't think they said it outright, but OpenAI has their funding coming in, and that clearly means that they're in a better position. Do you have any concerns here about their customer mix or diversity there?

22:27Tyler Radke:Yeah, I mean, I think when your biggest customer raises$110 billion, probably the largest raise for a company that I can think of, that helps a lot. um so you know the the alleviation of that that capital concern is a big one i mean look it's this is a lot of these contracts are are multi-year uh with with open ai they're ramping in 2028 29 and 30. uh so there's still um a lot of time left but look i think this market continues to be underestimated every quarter right if you look at industry projections on ai revenue capex they They continue to go up into the right. I mean, Anthropic is scaling at a massive rate.

23:12Tyler Radke:OpenAI is doing quite well, obviously, Google. So it's a big market. Capacity is increasingly in demand and there's a huge shortfall of that capacity. So we're very confident that the stuff that Oracle stands up is going to be eaten up regardless of its OpenAI or even some of the other hyperscaler or AI lab customers. Great.

23:36Stephanie Palazzolo:Well, Tyler, I want to thank you for coming on. That is Tyler Radke from Citi here on TI TV. OpenAI's revised ambitions in shopping have put a renewed focus on the ChatGPT apps that the company has enabled in collaboration with other companies like Uber and Instacart. But the first step to getting these apps to catch on is getting people to find these apps in the first place and making sure that merchants see the benefit of investing in these distribution channels. My colleague Anne Guillen wrote an exclusive story for the information on this exact topic today, and I want to bring her on to tell us more about her reporting.

24:14Stephanie Palazzolo:Anne, can you remind us what exactly are these ChatGPD apps? Have you used them? Sure.

24:21Akash Pasricha:So, yeah, it's interesting that OpenAI has chosen this to be the venue for their commerce ambitions for now, because I think, like you, Akash, many people haven't experimented with any of the apps in ChatGPT or even know how to find them or how to use them. Um, so right now you have to go to chatgpt.com slash apps and, uh, go, you know, you can browse through all of the apps that they offer and you do have to enable them. And then when you're in a chat, you do have to tag them or summon them. So if you're looking at recipes and then you decide, okay, I need to order groceries to make this recipe, then you would have to put at Instacart in your query to be able to call up the Instacart app.

25:11Stephanie Palazzolo:And then you call up the app, and then if you use the Instacart app, then let's incorporate our, the last time you were on, we were talking about checkout. So then with the app, does the checkout, does it take you to the Instacart website inevitably to do the checkout right now?

25:29Akash Pasricha:So Instacart is actually the one kind of example we have right now of where OpenAI's commerce strategy is going. So OpenAI is taking, instead of having the checkouts be directly in the chat window like they were before, they are now moving inside of these apps. So Instacart already has the checkout feature enabled in their app. So you would say at Instacart, I want to make tacos for dinner tonight. Instacart would call up a cart inside its ChatGVT app with all of the ingredients that it thinks you need. You can then go in and edit the store, what exactly you're ordering, that it will have your payment information from your Instacart account, and then you can check out, you know, right in that app interface.

26:24Akash Pasricha:And so that is, that experience is what OpenAI says is coming for other apps. So we will have to see, you know, like you mentioned, how many merchants are willing to invest time in building those apps. And then also, you know, if that's something that consumers start to use and start to respond to.

26:45Stephanie Palazzolo:And what do we know about how much traction these apps have gotten? I mean, you talked about Instacart. Instacart, I have to imagine that Fiji Simo is very invested in getting the Instacart work to app, given that she was the CEO of Instacart. But there are other options there. How much traction have these suite of apps gotten altogether?

27:05Akash Pasricha:Yeah. So when I was reporting the story, I chatted with a couple of companies that have worked on ChatGPT apps. And, you know, I think they all kind of acknowledge that they've seen pretty low overall traffic coming from these apps so far. But I did talk to someone who was a little bit more bullish on this concept, maybe than some others. And they said that, you know, while they were, it was fairly low traffic and it was a pretty small base. it has been growing over the past couple of weeks and months since they launched it late last year. So, you know, that could be a sign that consumers and ChatGPT users are starting to kind of catch on that these exist.

27:53Akash Pasricha:But I think that's an open question right now. and something that came up when I reported the story this morning was it seems like ChatGPT and OpenAI are still kind of weighing whether or not to keep the current setup where you have to explicitly call the app. They're debating whether it might be easier to, you know, suggest apps to users or kind of prompt them to use them in some way. But it sounds like that is still kind of being worked out.

28:26Stephanie Palazzolo:So do the merchants like them? And you talked to some folks for the story who observed this ecosystem at large and traffic. I wonder from the merchants end, you know, we've talked a little bit about the incentives for them participating in this. We've talked to the bookings.com CEO and asked him how much traffic he's seeing through this. He said, well, it's still very early days. And so I imagine traffic is not really the thing that they're measuring success on just yet? Like, how are they thinking about it? For sure. I mean, I think, you know, any of these commerce features that OpenAI has

29:05Akash Pasricha:experimented with so far, I think companies have seen them as experiments. And so it hasn't yet translated to a significant source of traffic or sales. And so I do wonder if because the app's approach shifts more of the work and more of kind of the development that's required, it shifts that more on to the merchants. So I'm curious to see if they think that is something that's worthwhile. A good example is Etsy. They are, the company confirmed to me that they are working on an app for ChatGPT. So we'll be interested to see how quickly they can get that up and running. but then you know the other side of the coin is shopify which although they're not a retailer or a marketplace themselves they obviously have millions of merchants that use their software to run their stores and shopify is not building an app for chat gbt they're just going to you know continue to have the product listings uh that that show up in search results that were there before but now to check out you will just have to go to whatever merchants shopify site to complete your purchase.

30:15Akash Pasricha:So, you know, you could lose some users along the way with that extra step to get to the checkout. But I think that approach is good for Shopify from their perspective, because they don't really have to do any extra work and the checkout still lives on their software.

30:33Stephanie Palazzolo:And very quickly, you also had some reporting about organizationally what is going on within OpenAI on the team that works on this? What did you find?

30:43Akash Pasricha:Yeah. So we reported that some of the teams working on this commerce effort, they were also working on other things. So for example, some people working on payments in the checkout system were also working on subscription billing for OpenAI's ChatGPT subscriptions. But the company did tell me that they're continuing to invest in commerce and they're expanding the size of the commerce team. They didn't give me too many specifics on what that looks like, but they say they're continuing to invest in building out that team.

31:22Stephanie Palazzolo:And I just want to make sure that I'm clear on one detail here because you've been covering this story from every angle. So the apps currently, the checkout itself, the checkout, your previous reporting told us that OpenAI was sort of moving away from having the entire checkout experience happen in the platform. Do the apps, do they have a checkout right now? And is that onus on Instacart right now? Or is it on OpenAI? Just help us square the reality.

31:52Akash Pasricha:Yeah, it's a little confusing. So Instacart is the example of the model that OpenAI is moving towards. So it will be an app where you can check out within the ChatGPT app. But remember, right now you have to do the extra work of calling up the Instacart app, logging into your Instacart account, getting to the checkout. So it is adding several more steps than before when you could search, you know, I want to buy a new sweater. ChatGPT would show you sweater results. you could click and then buy right there in the chat. So they are adding a little, they are adding more steps and definitely shifting more of that work onto the merchants and the marketplaces.

32:42Akash Pasricha:So I think that is maybe where the approach is a bit confusing. But OpenAI is saying this is in response to feedback from merchants and from users. So I think now the question is, what is the feedback on this model? Great.

33:00Stephanie Palazzolo:Well, Anne, I want to thank you for coming on. That is Anne Guillen, our e-commerce reporter here at The Information. The Information published exclusive reporting that OpenAI is planning to launch its Sora AI-generated video capabilities in ChatGPT. I want to bring on Stephanie Palazzolo, one of the reporters behind that story, to walk us through what she knows. Stephanie, tell us about your reporting.

33:24Ann Gehan:Thanks, Akash. Yeah, so as my colleague Sheree and I reported last night, and as you just mentioned, OpenAI is essentially moving this ability to use AI to generate video into its chat GBT product. And I guess kind of just taking a step back, so far this ability has been accessible through the Sora app, which is a separate app that it released last fall, about five or so months ago, where it kind of looks like a TikTok basically. But you can scroll through, you know, other users' AI-generated videos and generate videos yourself and share them with friends and on this kind of social feed. And so that has been kind of the way so far that people have been accessing kind of video generation capabilities.

34:11Ann Gehan:But now OpenAI wants to, you know, also add that into its ChatGPT product as well.

34:16Stephanie Palazzolo:So when I go into ChatGPT, I'm going to see ChatGPT, the text box, and then I'm also going to have a feed of all these short form AI generated videos that I can scroll through. Is that the idea?

34:27Ann Gehan:So we don't know the exact details of kind of like what the user interface is going to look like. I mean, right now, all we know for sure is that you will be able to generate videos through ChatGPT. We're not sure yet whether that kind of scrolling video feed is also going to be coming over. However, you know, we have reporting that the kind of social app will continue operating, but it's not clear if that is kind of the long term plan for the app or if at some point it's going to kind of end up shutting down basically.

34:58Stephanie Palazzolo:But it is kind of an interesting question to think about. And we've seen OpenAI pivot in other parts of its consumer app ecosystem here in recent weeks. The idea that we don't know what will happen to the social aspect of OpenAI's product portfolio, because that's what Sora was supposed to be. is it was, I mean, there was even a discussion at one point about you being able to share, was it, well, not sharing chat GPT results, but I mean, there was a whole play here around being able to make chat GPT a more collaborative experience. And so if it's really right now, if all we know is that it's the video generation, then that's certainly something to note here.

35:38Stephanie Palazzolo:I want to ask you about, look, does this relate to the code red that we saw a couple weeks or months ago from Sam Altman saying we have to get people back into ChatGPT?

35:49Ann Gehan:Yeah, so it's not clear, I guess, whether this is, you know, kind of explicitly part of Code Red, which actually, you know, ended, you know, earlier this month. But it definitely is, like, kind of in the spirit of this Code Red. So basically, we've kind of seen from OpenAI, I feel like the kind of stereotype of OpenAI for the last couple of years is that it's very experimental it seems like researchers like individual researchers and engineers and product people just kind of come up with ideas and like put them out into the world which has been like really great i think has led to a lot of great products but it does feel maybe at times a little bit like all over the place so you know since uh last fall the company you know has gone under this kind of code red uh state where basically it just really wants to you know focus and and lock in on its main money-generating product, which is ChatGPT.

36:40Ann Gehan:So this does kind of feel like an extension of that, where it's like, you know, why even have this, like, separate app where people go to make videos? Like, that should be in ChatGPT. That should be kind of a draw to ChatGPT.

36:52Akash Pasricha:Right.

36:53Ann Gehan:And so it does kind of, you know, feel like OpenAI is trying to overall, like, simplify and just really just, you know, kind of bring everything together product-wise into something that feels a little bit more cohesive and centered around the ChatTBT product.

37:10Stephanie Palazzolo:Man, I did not realize Code Red was in the fall. Gosh, it feels like it was literally just a couple weeks ago. I don't know, time kind of blends. Yeah, very much blends together.

37:22Ann Gehan:I think it technically started in December. So I guess winter would be a little better. But I think originally people thought it would be like a couple week long thing and it stretched into a couple months. So I think it took longer than what most people expected, but it is over now, it seems.

37:39Stephanie Palazzolo:Okay, so let me ask you about the financial aspect of this. You pointed out in your story that this could have big repercussions for the company's inference costs and how much it costs them to allow people to generate these videos. I understand that part. I am sort of curious how you are thinking about how this might help them generate more revenue. I mean, it wasn't really ever clear to us how they planned to make money from Sora. If it was a social media feed, then, you know, was there ever going to be an advertising play? I mean, do you have any ideas on how you think they could squeeze more revenue out of customers with this type of functionality?

38:19Ann Gehan:Yeah, so, I mean, this is pure, you know, speculation. Speculate.

38:23Stephanie Palazzolo:I'm just, I'm brainstorming here. I'm a brainstorming. You know, maybe this is more premium subscriptions or, you know, maybe that's, it's, you know, we've seen you can pay instead of$20 a month, maybe it's$7 a month for the Sora capability or something like that.

38:39Ann Gehan:Yeah, exactly. So I think this could be a draw to more premium subscriptions of ChatGVT. It could be something more like a usage-based model where maybe if you're a filmmaker or a creative, you're paying per video that you generate. So I'm sure they're thinking of ideas of how to kind of like eke more sales out of this. But as you mentioned, I think the bigger thing that Tree and I were kind of thinking about is what this means for costs. Just because, you know, a year ago, there was this big viral moment where everybody was making kind of like anime versions of themselves, basically, with ChatGVT whenever they first came out with, you know, image generation capabilities.

39:22Ann Gehan:And that really pushed Open AI is kind of their like compute to the brink. And so, you know, video generation is even more compute intensive than image generation. So, I mean, this is just going to further kind of put pressure on the compute resources that Open AI has and could, you know, definitely be a driving force behind higher costs. Right.

39:49Stephanie Palazzolo:Yeah, no, it is. It's interesting to think about for sure. And again, to be clear, we don't know how they plan to make money out of this, but it is sort of where my mind goes. You know, how could they slice this product in different ways to get more revenue out of it in the end? So, Steph, it was a great story. I want to thank you for coming on. That is Stephanie Palazzolo, our AI reporter here at The Information. Our next segment is with our sponsor, PwC. Software sales teams have had to reinvent the ways in which they pitch their products to customers in the face of AI. Although the truth of the matter is software sales were already changing quickly before AI became popular with trends like consumption-based pricing dominating the conversation.

40:34Stephanie Palazzolo:I want to bring on Dallas Dolan, a leader at PwC's tech media and telco practice, to help us understand what this looks like in practice on the ground. Dallas, welcome back to the show. It's great to have you here. Hey, Akash. Great to see you. So when you think about tech companies selling AI to businesses, how does this pitch look today? And how does that pitch look different than selling software five to seven years ago?

41:00Dallas Dolen:I think maybe one way to frame it out is, you know, as we talked about last time, you know, SaaS is still here, but it's the SaaS 2.0. And that's what is enhanced by, you know, AI or gentrification. The biggest change that I think the customers are demanding, and at the end of the day, they're the ones, you know, fronting the bill, right? They're writing the check. It's all about what are the outcomes and how do I get value from that, which I am signing up for. And that's a very different concept, if we're being honest. If you look back to the world of ERP over the last 25 years, certainly the world of CRM over the last 10 or 15, the integration of AI is going to be one where people are expecting and have been told and promised, candidly, they're going to get a lot more value, a lot more experience, a lot more, frankly, perhaps it's scalability out of all these things.

41:52Dallas Dolen:And the only way to really prove that that's happening is to move to something that's a lot more outcomes-based, which is, candidly, like you said, where companies have already gone. If you look back at history here, even in the last few years, you have examples of it. Autodesk just went through this. You have certainly Adobe from a number of years ago to readjust the way that they go about selling the creative suite. So there's examples of companies who've gone there. And I think just the next generation of sales team, as well as maybe it's sales leadership and even accountability framework. That's really what the direction of travel tells me is going to have to happen in the market for other companies to be successful to follow those same journeys.

42:34Stephanie Palazzolo:You talked about outcomes here, And this sort of leads to the question around how we measure ROI for these AI investments. How do you see your clients evaluating that question today? And how does that compare and contrast to the way they evaluated ROI on the software that they spent on five to seven years ago, sort of when the SaaS movement was really starting to take off maybe a decade ago?

43:02Dallas Dolen:Yeah, I mean, it's still in the vernacular, right? If you look at the way that we think about system integrators, it's all about the go live. It's the did you put the system in and did it like turn on? Were people able to access and do the things that they were supposed to do? That generational or transformational dynamic tended to be still elusive. I mean, it wasn't that you couldn't find some efficiencies, but broadly speaking, it was all about getting to a fixed date or maybe it was a slip date, but still a fixed date that might have been a little later that made you happy because you got the ERP in or you're able to migrate, let's say, from on -prem.

43:35Dallas Dolen:cloud. This new thought process, if you will, from an actual outcomes view of the world, I think it's two bits. One of them is how much money was spent over the last couple of years on AI, especially from a POC point of view. We were measuring value on a, well, how many POCs or how many pilots are you running? And honestly, I mean, that's a fun metric, but at the end of the day, that's not giving you really any sense of outcome. That's not, you know, conjuring up new customers. It's not making your process more efficient. It's not allowing you to scale, perhaps, you know, in engineering or scale and in the front office the way you might want to.

44:12Dallas Dolen:And so that's why this outcome concept, I think it's actually coming for all of SaaS, including, you know, ERP world, but it's really coming into the fray as it relates to AI. And it's going to, you know, probably also begin to show us like who the winners and losers are, right? I mean, you will have the data-rich companies who have the ability to monetize much more rapidly and also be able to extrapolate on that concept of value much more easily. And you have those that are more superficial, perhaps, that will struggle with this refactoring, for lack of a better word, this move to 2.0. I actually spent last night with about 110 sellers of AI products from one of the large companies here in the Valley.

44:57Dallas Dolen:I can tell you that a lot of them are actually excitable about the concept of being able to prove the value proposition, but they're also, you know, struggling up against the reality of, hey, I'm still accountable for a number. And that number is how many licenses did you sell? And that's going to be the big friction point probably throughout 26. And we'll see how things go as it relates to, you know, that migration, but it's going to be a challenging one for people to move between.

45:21Stephanie Palazzolo:But I want to go back to this idea of the go live being the success metric. Maybe in the, I don't want to call it the older era of SaaS, but the pre-AI era of SaaS. I mean, why was that accepted as the metric for success back then? And why weren't they asking sort of the fundamental questions they are right now about how many dollars are we saving? What is the ROI?

45:48Dallas Dolen:Yeah, I think it's two things. If you consider this fact, right, if you look at CEO interviews, and even commentary on earnings calls, go back just a few years, no one with unless there was a major, you know, incident or problem, if you will, from an implementation of software point of view, ever talked about anything that was going on in the back office or the front office or, you know, the supply chain as it relates to technological implementations. implementations. Now you have it, you know, and people are running accounts on these calls, right? You have CEOs and CFOs talking about AI, right?

46:24Dallas Dolen:Again, the new SaaS 20 times, 30 times, 50 times on a call. At the same time, you know, when they're surveyed, and we did this just a few months ago, these same executives are saying, hey, we're not seeing the outcomes from it. So I think it's a direct accountability all the way up to the top of the house. It's actually changing the dynamic. Whereas once before it was perhaps a bonus or a stock award for a CIO, it's now becoming, you know, a board level issue that is being talked about on a day to day basis. And really, you know, again, back to that accountability point for, you know, the executives themselves, if it's on their front burner, they are going to talk about it.

47:00Dallas Dolen:They're also going to help, you know, hold more people accountable to the same.

47:04Stephanie Palazzolo:So if it's if it's now a CEO level issue and a board level issue, does that make the investments go faster? Does that add more layers of decision making that have to happen? What's the net impact of that on these AI investments?

47:21Dallas Dolen:Yeah, that's a great question. I think it's going to put a lot more pressure downward back into the organization on, you know, how do I allocate resources, both time and money? And it's also going to put a lot. And this is where the sophistication of management and boards is going to come into play. We talked about this quite a bit recently with the number of boards that we meet with. And that is, you know, do boards have the ability to hold management accountable, you know, on their goals and on the outputs that they're looking for? Presently, it varies by industry, I guess, you know, nicely put that.

47:55Dallas Dolen:But the reality is it's harder to say exactly how that one's going to play out. I think you're just going to see a lot more pressure, you know, downward as it relates to the how much money is being spent on it without being able to demonstrate ROI. One thing that we've seen is a lot of companies asking, saying, hey, what's the best way for me to actually capture and report ROI? Right. The CFOs want to bring it. Procurement organizations want to bring it. Certainly, you know, the CEOs want to see. And I think at some level, the boards are going to be asking for it as well. So you have this upward, you know, dynamic coming through and pushing hard to, you know, to allow for the reporting to happen so that that accountability can happen, too.

48:32Dallas Dolen:And it is all about, you know, what gets counted gets done. I think that applies absolutely in the case of, you know, again, this new age SaaS with AI embedded in it.

48:42Stephanie Palazzolo:Great. Well, Dallas, I want to thank you for coming on. That is Dallas Dolan from PwC here on TITB. That does it for today's show. Oh, a reminder, we are on this stream Monday through Friday at 10 a.m. Pacific, 1 p.m. Eastern. I want to thank you all for tuning in. We really do appreciate your viewership. Make sure to subscribe to the information on YouTube and follow us on X, Instagram, TikTok, and check us out wherever you get your podcasts. I'm already excited for our next show tomorrow. Have a great rest of your Wednesday. Bye-bye for now.