Apple-F1 Media Deal, Moneyball for Minors, Oracle's AI Margin Worries, AI Rollup Boom | Oct 17, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Basricha. It is Friday, October 17th. We have got a jam-packed show for you today. We are breaking down everything you need to know about Oracle's new revenue projections it unveiled for analysts yesterday. But also, we have some new reporting for you that gives even more context to the company's margin structure. We're then going to shift into a discussion about the business of AI roll-ups, which are becoming a very popular strategy in Silicon Valley right now. And finally, we are bringing on our sports business reporter to talk about the new Apple F1 deal and about her feature story on the booming business of technology that is helping people stream youth sports.

0:56We've got a lot to get to, so let's get right on into things. Oracle held its analyst day yesterday, and one of its big topics that management tried to give more context around is the company's long-term profit margins. The information, of course, has done a ton of reporting on the current state of profitability of Oracle's AI cloud business. And yesterday, my colleague published some new undisclosed details about its cloud business that offer even more context. I want to bring on Anissa Gardizi, our cloud and compute reporter, to talk to us all about what she learned yesterday. Anissa, it's great to have you back on the show.

1:29Welcome. Hey, gosh. So big day yesterday for Oracle, a lot of new information. The big headline that I think everyone was paying attention to, of course, was the revenue projections that it lifted. Tell us a little bit more about some of your other takeaways from the analyst day, and then we'll get into your reporting in just a second. One thing that Oracle executives wanted to talk to investors about was the gross margins of its cloud business and all of the different segments that make up the overall gross margin that they will be shooting for in the future. and the new co-CEO Clay broke out the gross margins for each of these businesses.

2:09And everyone was really zeroed in to the gross margin for the AI infrastructure business. This is the business where Oracle signs big contracts with Meta, XAI, OpenAI. And it has to fill data centers with extensive NVIDIA chips and then rent them out. And this is much lower margin than the other Oracle Cloud business lines. And so what Clay said was that they're shooting for a gross margin of between 30 and 40 percent, which includes the ramp up period where they are filling a data center with chips and maybe not getting revenue yet. And so that was news to investors that they were shooting to 30 to 40 percent.

2:47But like you mentioned, our historical data that we were able to find has them below that. So that's definitely where management is looking to get. But we sort of went into deeper detail on how it's been going in the past year. And tell us about this ramp up period that executives talked about. Yeah, it's a really important moment as you build out your data center fleet. So it shouldn't be overlooked. It is important. Essentially, Oracle works with partners that build its data centers. And Oracle doesn't pay them until the data centers are ready. But once those data centers get handed over to Oracle, it's on them to fill it with all the chips before they can hand it over to customers.

3:28And so there is a period of time executives talked about where they are working on getting a data center ready for their customer. They're incurring expenses before revenue starts coming in. And what management really tried to hammer home was that that ramp up period is not going to be a significant cost, especially if you factor in maybe the next quarter or the quarter after that of revenue being higher than expenses. And so I think that's really one thing that they spent a lot of time on, getting people comfortable with this new business line that Oracle historically hasn't really played in.

4:01Okay, so that's all the things that are some of the things that Oracle talked about at the analyst day. Talk about now some of the documents that you got a hold of yesterday that give us some more context about the margin structure of the company. Yeah, of course. So last week, we reported on the overall gross margin in the past quarters being on average across five quarters around 16%. And, you know, one thing that was important to note there was that that included the ramp up period of the Blackwell chips, which, or, you know, OpenAI hasn't really started paying Oracle that much money. And so the margins look negative on that chip.

4:40So one thing we wanted to do in the story yesterday was say, okay, even if you take out the ramp up of the Blackwall series chips, how is the rest of the AI infrastructure business doing? Because they have A100 chips, which came out before the AI boom. And then they also have the H200s, the H100s. And so we provided more information on in the last quarter, how did those different chips line up on a margin perspective? and they were all under the 30 to 40 % gross margin range. Of course, this is a point in time, so important to take that into consideration. But these are older chips that they've been using for a while, is the point.

5:19Exactly. These are chips that they've had in their data centers for years. Right. And, you know, they do have customer contracts. And so one thing we also saw with the H200 was that they had pretty low utilization of that chip until last quarter when they signed a new deal with OpenAI, which improved the margin to 19%. So I think it's just, you know, this data is, of course, a point in time, only a couple months, and Oracle is growing. But they're really interesting data points to understand, can they hit the 30 % to 40 %? One of the things I found interesting about that chart, though, that you were talking about was that, I mean, this also included non-NVIDIA chips.

5:56I mean, it showed the margin structure for other companies' chips. I mean, what have you heard about how they're dealing with sort of diversifying beyond just NVIDIA. Yeah, I mean, Oracle wants to get more AMD chips into their data centers so that they have more options for customers, but they're in the very, very early stages of having them deployed at scale. So I think the numbers on AMD reflect, you know, a small number of chips. But I was talking to people yesterday in Vegas about AMD chips and Oracle data centers, and I think it's something that they really want to do. And I don't think Oracle is going to take a big risk on AMD chips, I think they're going to structure these deals in a way that makes sense for them long term.

6:34Right. Well, I'll tell you what, I want to continue this conversation and I want to bring on Derek Wood, who is an analyst at TD Cowan, who covers Oracle. He, of course, was covering the analyst day as well. Derek, it's great to have you on the show. Thanks for being here. Thanks for having me. So tell us, look, I want to talk about some of the reporting that Anissa did, but look, the stock is down today after all the reassurance that the company gave. Yeah, that it is. I mean, it was up 3 % yesterday on the news, and so it's giving back some of the gains and then a little some. When they did give annual guidance over the next five years, there was a little bit more back-end loaded EPS ramp in fiscal 29 and fiscal 30.

7:17Fiscal 28 EPS was a little below where Street was. So I think that may be a factor in seeing the layout of the trajectory of EPS. Now, I think part of that is they've embedded some incremental interest costs, assuming that they may raise more debt. And they threw just a lot of conservative assumptions in those numbers to have cover for various funding needs. But we still got to the path of fiscal 30,$21 in EPS, which was above street at$18.50. So I thought, you know, nice outperformance on the targets when you look at the fiscal 30. Right. What did you make of the discussion of the different business lines within the cloud business?

8:00Yeah, I mean, they talked about the AI infrastructure. 30 to 40 obviously addresses a big investor question. The cloud natives, 40 % to 60 % gross margin, average deal size of$97 million, and growth rate's very good in that. And then the enterprise could be anywhere from 60 % to 85%. They gave us database revenue targets, cloud database for$20 billion in fiscal 30. Big inflection starting to happen at fiscal 28. That business is very high margin, I think 80 % plus. So those are the three buckets we have to go with. Clearly, I think the AI revenue is the biggest inflecting piece of the model. And it was nice to hear that, you know, they gave us an example of 35 percent gross margin.

8:48Yeah. With a 60 billion dollar six year type of contract example. And you and they're booking 21 billion in and, you know, and total gross profit over that period. I think that's how they structure those contracts. Yes, there's going to be initial cost on the ramp in terms of having lower margin at the beginning. But I think when they look at a six-year contract, this is how they structure the profit outlook for Oracle. How much do you think about the things that the company didn't tell us yesterday? I mean, you talked, you know, it didn't tell us a lot about capital expenditures, for example, what it's going to have to invest to to sort of make all this cash flow positive in the long run.

9:28And we didn't hear too much about the size of each of these segments, even the time frame for that 30 % to 40 % gross margin. I mean, how do you think about those questions and some of the, I guess, mystery, I guess, about what we need to still wait and see? Yeah, I mean, the reason they didn't give CapEx is because they may do vendor financing. And that could come in the form of kind of interest costs to an NVIDIA or an AMD instead of CapEx on Oracle's P &L. And there's just different ways they may fund this that may not come in CapEx. They may get their data center operators, and this is something we've talked about, get them to purchase some of the equipment.

10:16And so it then becomes more of the lease payment, which is an OpEx for Oracle instead of a CapEx. So that's the reason why they didn't want to give specific numbers on CapEx, because it really could come in different forms. Right. And what about the time frame, the 30%, 40 %? I mean, we still don't really know what long-term really means, right? Yeah, this is true. in that example of$6 billion per year for six years, or$10 billion per year for six years. I mean, they did kind of show in year one already at like every year the same profit level. So was it like that to me made it seem like there's not going to be a long time to ramp to get to where they need.

11:01But they didn't give us any time frame, Keller. Yeah. And you said, if we go back to sort of some of the unanswered questions with respect to Oracle, what were you hoping to get a little bit more clarity on yesterday? Or what are you hoping to get more clarity on in the weeks and months to come? Well, I mean - Well, the gross margin and how much money they're making. I was gonna, Derek, let me come back. Let me go to Anissa first and then we'll come back to you. Oh, I was just gonna say that I think one thing that they did, they were sort of trying to talk about, you know, I was curious about the risks.

11:35So, you know, you can have like a plan to build a data center and you can have your schedule and you can sort of try to time your revenue to your expenses. But one thing that I'm definitely seeing in the industry is that timelines can slip and, you know, you might think that you're going to hit a certain date, but there are like unforeseen circumstances that happen. And so I think just as like the base number gets bigger on how many gigawatts Oracle is trying to big out to build out, I just wonder, you know, how much they're factoring in, you know, the chances of projects getting done exactly on time, because I think they really do care about lining up their revenue to expenses.

12:11Right. Okay, Derek, now you take us home. Well, we're looking for Abilene to really ramp at the beginning of next calendar year. And then for that to have another, the broader Stargate project they're involved with, what we think is going to produce$60 billion in annual revenue and$20 billion in annual profit by the time we get to fiscal 30. And there'll be a series of kind of build outs that happen across different data centers. And there'll be a mix of different costs in and out along that front. But, you know, we certainly think that by fiscal 30, it's going to take some time. We're going to have some really big revenue numbers out of Stargate.

12:58And we already heard from Oracle. They signed$65 billion in contract value in the last month and a half. It's not open AI. I think it's meta. We think there's a lot of other big deals for Oracle to go out there. It'd be incremental to our numbers. Great. Well, Anissa and Derek, I want to thank you both for coming on. It was certainly a jam-packed week of news for Oracle, and I think there's going to be more to come. Anissa, thank you for joining us. Derek, thank you for joining us as well. We look forward to having you on once again very soon. Okay, we've talked a lot on this show about the complicated ways that AI can help big enterprises use the technology.

13:32It turns out many people are focused on helping smaller businesses use these tools. And in fact, there is a new branch of venture capital that is focused on consolidating smaller businesses rolling them up into one and using AI to make them more efficient, effectively acting kind of like private equity does. I want to bring on Vinay Iyengar, the managing partner at Anansi Capital who has been watching this trend play out and is certainly diving headfirst into it. Vinay, welcome to the show. It's great to have you. Thank you so much, Akash. It's great to be here. So can you explain to us what this boom in AI rollups is all about?

14:05Sure. I mean, I think my contrarian take that I've been thinking about for a long time is that the traditional SaaS business model is dead. The business model that has taken up the lion's share of venture capital dollars over the past two decades is sort of coming to an end. And I think the next decade is going to be littered with a bunch of carcasses of these types of companies. And the reason for this is sort of multifold. First, we're entering an era where software is easier than ever to build. So historically, R &D was a big portion of the costs for a software business. And today, R &D is less and less of a relevant part of OPEX for a software business.

14:42Secondly, one of the underlying things that made SaaS businesses interesting was the fact that COGS were very low. It was largely just cloud hosting bills where you would pay AWS or GCP. And then as a business scaled, those COGS would ultimately be negligible, and these businesses would be very profitable. Today, we're entering an era where the actual costs of training and inference for these AI-enabled software companies is actually very high. And so these businesses are not operating at the same gross margins that historic SaaS businesses were able to. And so the traditional venture model of funding these AI-enabled agents or AI-enabled SaaS companies is no longer as tenable as it was five or 10 years ago.

15:23And so investors are looking for new places to deploy capital. And one of the places that I consider very interesting is this area of SMB consolidation or AI roll-ups, as some people call it, which is this idea of founders going out and buying these old school SMBs, old school services businesses. We're talking like legal clinics? What are we talking about here? Yeah, we're talking about all sorts of things. We're talking about businesses that make up, call it at least 50 % of the US GDP. These are small hardware stores. These are small clinics. These are insurance brokers, freight brokers, real estate brokers, property management companies.

16:01It really runs the gamut. And so the idea is that you buy a number of these companies, you smack them together, you effectively get rid of some of the redundant operations, I imagine. And then AI is used to do what exactly? What parts of it can be made much more efficient? Yeah, so it really depends on the business. But broadly, we think about AI as being a lever for either increasing the margin of this business, so basically automating large swaths of the cognitive labor that's being done in these businesses. And then in many cases, AI is also used as a lever to actually drive top-line growth and allow this business to grow a lot faster.

16:39How do you do that, though? So, again, it depends on the context. I'll give you a couple examples to just give you a sense. One of my platforms that I invested in recently is a roll-up of accounting firms. And the idea is AI can actually help an accountant serve more customers per year than they might have been able to do historically. Let's call it double the output of a traditional accountant. And in doing so, that accounting firm is now able to serve sort of 2x the number of customers. And so AI in one case is a means for just increasing capacity and being able to service more customers in your competition, which ultimately drives top line growth.

17:18Another example, I'm seeing a bunch of really interesting cases where founders in places like roofing, for example, a roll up of roofing companies are able to actually use Google Earth data to figure out what are the roofs that have a lot of leaves on them in a certain locale. And let's go target those companies. Let's build a AI agent that basically is able to automatically go figure out what are these businesses or who to go after, essentially. Who do I go after? Let's automate the actual outreach, right? Okay. Actually send them an email automatically. Let's automate everything to do with intake and scheduling.

17:53So AI is really automating broad swaths of the admin labor of these small businesses that historically took up, call it, 20, 30 % of the operating expenses. and is now allowing these businesses to really just focus on their core business, which is serving their customers. So do you consider yourself a venture capitalist or a private equity investor? It's a great question. I consider what I do to be similar to venture. Sounds like private equity to me. It sounds like private equity. I think the rose-colored way to think about what I do is it's venture-like upsides with private equity-like downside.

18:27But the fundamental job that makes an investor like me good or bad, which I'll see over the next decade, is can I properly identify the right sectors? And can I properly identify the right founders to actually go out and execute the strategy? And so unlike a few of my sort of competitor firms out there, which are building truly what look like lower middle market PE funds, they're actually going out and executing the roll-ups in the digital transformation themselves. My strategy is to actually back founders, who I believe are really talented and hungry and get out of their way like any great VC should do and let them actually execute both the M &A and the digital transformation.

19:05So the strategy that I'm employing, that firms like GC are employing to an extent, look more like venture and whether or not we're successful or not will really depend on underwriting amazing founders. Let me ask you one last question before I let you go. And maybe this relates to what you were saying earlier at the start. I'm wondering what competitive advantage a firm like you has, where you look to consolidate these businesses, build out software that can help them make things more efficient. What is the advantage that that strategy has compared to a software company that says, hey, I'm going to only focus on the roofing companies and I'm going to build the most specialized software for every roofing company in America.

19:46We've seen that model play out with, I mean, I'm thinking about fitness software companies, for example, why would they not just go to that software company to lower their costs that way? Yeah, it's a great question. And I think part of what you're asking is why is there alpha in actually rolling these companies up versus actually just building vertical software and selling into these? Yeah. Right. And my thesis sort of goes back to what I was saying earlier is we're entering an era where software is easier than ever to build. And so in any given category of vertical software, there are now a dozen competitors that have reached feature parity and call it three months.

20:24And if you don't believe me, you just look at the latest class of YC companies, you'll see a dozen AI-enabled agents in each subcategory of SMB software. And so competition is higher than ever. CAC is higher than ever. And so I find it very hard to believe that we're going to see multi-billion dollar exits in these sort of categories because you're just littered with competition. As I was alluding to earlier, the business model doesn't necessarily make sense anymore when the cost to actually train and do inference for these models is so high. And so my thesis, and I think a lot of people are slowly catching on to this, is that rather than bet on one of these 12 AI point solutions, why don't I just own the customer?

21:09That to me makes the most sense as the right model to play. And in many cases, the platforms that we back, rather than build these software tools in-house to increase margin and drive growth, they're actually just buying the off-the-shelf software. And I joke with my founders, I call it taking the VC subsidy. You have a bunch of VCs that have put in hundreds of millions of dollars into these vertical software businesses, and they're competing to win customers, so they're pricing really low. And so for me, as a customer, I'm able to buy these tools very cheaply off the shelf. I don't need to reinvent the wheel.

21:41And so that for now is the strategy that we've been employing. Great. Well, thank you, Vinay, for coming on. It's a fascinating strategy. It's one that we've been tracking very closely here at The Information. We will link some of our reporting in the show notes. That is Vinay Iyengar, the founder and managing partner of Anansi Capital here on TITV. Okay. There was big news in the sports media rights world today. Apple and Formula One announced a new five-year partnership that will bring F1 races exclusively to Apple TV in the U.S. starting next year. I want to bring on our sports reporter, Sarah Germano, and our features editor, Nick Wingfield, to help us break this all down.

22:17Sarah and Nick, welcome to the show. It's great to have you. Hey, guys. Okay, so let's talk F1. Sarah, what did you think of the deal? Well, there's a couple different ways to slice this, right? Not to use an Apple pun. But if you're talking about pure dollars, it's obviously not comparable to some of the other major league sports in this country. You know, NBA and the NFL are getting billions of dollars per year from a variety of media partners, from linear cable and streaming. In this deal, F1 is going exclusively to Apple. It's more comparable to one of Apple's existing deals with MLS, where all their games are going in one spot.

22:56And they're almost doubling what they were getting from their current broadcast partner, ESPN. They're going from about$85 million a year to$140 million a year. So it's a big check for F1. But if you're a consumer, on the one hand, great news. You can go and get all F1 races in one spot. If you're not already subscribing to Apple, it gives you a reason to. And it's not that big. I mean, you were saying this. It's not that big a deal compared to other deals. Like even NASCAR is much bigger, right? Yeah, there's a lot of sports. And again, it's comparing apples to oranges when you think about the viewership that each of these sports get, the length of their seasons, and even just the number of broadcast partners that they have.

23:39But one thing I've been hearing consistently when I talk to people in the media and like the sports league worlds, if you have a sport that has a global fan base, which F1 does, the streamers are very interested in having sports with global fan bases because you don't have to deal with media partners that may be just domestic and cut up a million different deals depending on what geography you're in. So it makes logical sense why a sport like F1 would go to a global tech platform like Apple. Nick, as you've seen Apple's media strategy play out, we mentioned MLS. What did you make of them going after F1 and what does it tell us about their strategy broadly?

24:21I guess it's the least surprising deal in the history of deals given that they, you know, they'd been rumored to be in talks with them and they'd also produced the F1 movie, which is apparently the biggest gross. Good movie. It's a good movie. Yeah, you know, I think their media strategy is a bit puzzling. You know, on the non-sports side, they've really kind of positioned themselves in some ways as the HBO of the streamers out there. They tend to do like a smaller selection of, you know, different types of entertainment, some successful, some not. And with sports, you know, it's not attempting to be comprehensive, but it's a nice benefit.

25:05You know, I am sort of curious about the demographics of F1. That audience may skew a little bit more affluent, which is kind of part of Apple's brand. So as a consumer, it's great. I mean, especially for cable cord cutters like me who no longer have access to ESPN. Now I can get access to all of the F1 sports through my Apple TV subscription. Remember, it's Apple TV. It's not Apple TV+. Right. I should say that I did Google it today and they haven't yet really updated the website with the getting rid of the plus, which I thought would be the easiest part of this exercise. But I think they're a little slow to the gate on that one.

25:51Well, I'll tell you what, it's certainly an interesting deal. Stick around because, Sarah, I want to talk about your story in a second because you had a very interesting one. So we'll see you in just a second. I do want to talk about our Weekend Magazine feature. There is a massive untapped market for investors in the sports world, and it's not just professional or collegiate level sports. It is youth sports. The average American sports family now spends more than$1 ,000 a year on their child's primary sport, a 46 % jump since 2019. That is according to the Aspen Institute. Much of that is fueling a boom in tech tools, everything from stat tracking to live streaming games.