Inside Sora 2, Fast Money in AI, Fermi America’s IPO, Outdoor Voices’ New Collection | Oct 2, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Pasricha. It is Thursday, October 2nd. We have got a great show planned for you today. We are unpacking the launch of OpenAI Sora 2 with Bain Capital Ventures. We are also talking about one of the craziest data center companies that went public this week, and separately, about some of the reality checks that may come for AI startups whose revenues have skyrocketed overnight. We've then got one of Sullivan and Cromwell's top M &A lawyers joining us to talk about the current state of deals in technology. And finally, Ty Haney, a friend of the show, is going to come on to tell us more about Outdoor Voice's new product line launching today.

0:55Before we get going, I want to highlight for you a story that we just published this morning at The Information about Meta Super Intelligence Labs, also known as MSL. The story is a deep dive into some of the frustrations that have been brewing on the research team within MSL. And a key detail here is that researchers have become frustrated about a new policy that says that their work will need to go through additional reviews before they can publish it. We will link that story in the show notes. I highly suggest you check it out. And with that, let's get to our first guest. This week, OpenAI launched Sora 2, its new video model that has been the talk of the town.

1:35I've had many friends sending me AI videos that they have generated all week. Many of them have been quite impressive. But the social media angle of the Sora release is equally captivating people right now. And I want to bring on someone who wrote a good piece on that. Sonia Oja is a partner at Bain Capital Ventures. It is her first time on the show. Sonia, welcome to TITV. It's great to How are you? Akash, lovely to be here. Long time watcher, first time on here. So let's get into it. We'll have to make it long time watcher and many time guest. How about that? I'm really excited for this conversation.

2:10Very quickly, Sora, what have you been generating with it this week? Anything? You know, I have been mostly perusing, but that is not too different from my use of other social media. I'm alert. But I have been very entertained, I will say. Great, great. Well, there's a lot to get to. You wrote a great piece about Sora 2 and particularly about the launch strategy. Talk to me about what really stood out to you, not just about the model, but the way that they actually rolled out this product. Yeah, absolutely. So it is interesting. I feel like for a while now we've been talking about what is AI native social.

2:46There have been rumors about OpenAI releasing something like this. So I feel like the world was prepared. but what was fascinating is how decisive they were with this launch and how many opinionated takes they had in how they rolled things out. So the first thing, obviously, you mentioned it already, everyone was hankering for these invite codes yesterday. It's a stroke of brilliance for two reasons. One, this playbook is as old as the Valley. Gmail started off with these invite codes. I don't know if you remember. You know, they used to sell - It was always superhuman for me. That was the invite quote I was trying to get.

3:25That makes sense. You know, whatever it is, the virality comes from the FOMO, comes from the scarcity. We know this playbook. But what was interesting in this situation was that it doubled up as a throttle on chaos. Because when you're putting out a generative video model into the world that's as good as Sora, there's going to be deep fakes, there's going to be abuse of the platform, There's going to be a thousand different ways people will try to break it. Doing it like this, they were able to, you know, reduce the flood on themselves as they sort of waded through it and set standards as they went along.

4:00So I thought that was really smart on both levels. The same thing they did, which raised a lot of eyebrows, caused a lot of consternation, and I don't think we've heard the end of it, is their stance on copyright. Opt-in by default is bold. You can expect to hear a lot from entertainment lawyers, all the production houses. What is fascinating is that this is coming so soon after the Tilly Norwood fiasco that happened in the media. And you're seeing people get up in arms, actors, guilds issuing statements about how this is not the way forward and creativity should be the domain of humans. And for those of us, I want to sort of translate the opt-in part of this.

4:45When you say opt-in, what does this actually tactically look like for actors and people in the entertainment industry? Yeah, it means every piece of IP that is online can be used by OpenAI and Sora as a part of the generation unless the entity actively opts out. So you have a character, a movie, a song, that's fair game unless you send them an opt-out notice. So conceivably this right here, I mean, you know, as soon as it's posted on YouTube, this is fair game is what they're saying. Yeah, that's what it sounds like. And I'm sure this will get battle tested in the days to come. But the other one that's raising eyebrows is that you can't do a blanket takedown.

5:23So Disney can't come in and say, hey, don't use anything that we've ever done. They have to do piece by piece. They have to go, you know, IP by IP and say, you can't use this specific movie or character. And they have to do it a thousand times over. So it's essentially like you're draining the water from a sinking ship using like a thimble. Just keep going. Death by a thousand cuts. Right. And it makes sense because from one perspective, it scales up out, does not scale. You have to go, you know, hanging to people asking for permission. Opt in. It's everyone else's problem. So I don't know how this is going to withstand legal pushback, but we'll see.

6:04If it does withstand legal pushback, they're going to set a precedent for this industry. and as it languishes in the courts, they're going to sort of set a precedent for consumer behavior and expectations around what goes on in this. So I'm curious to see how this stands the test of time. So do you think that this is good for the world? Because I was struck by sort of the bottom half of your newsletter, which was really, it was your take on whether or not this is a net good for the internet and for the world. And I just want to read a passage. You said, when reality fragments into infinite simulations, we don't just lose common ground.

6:43We lose the ability to argue over it. If everyone gets a custom show tailored for just them, we're not sharing culture anymore. So what's your take on that? Was this good? Well, I think it was inevitable. I think in its current form, I'm not a consumer investor, but I don't see how this is going to scale. because I think once the novelty wears off, you don't want to see your friends riding dragons. You kind of want to see what they had for dinner and what they're doing. Or you want to laugh at other human beings. Maybe they've slipped on a banana peel on TikTok, stuff like that. I do think that it's going to adapt and it's going to change with how consumer demand is changing.

7:27But I stand by it. I do think it's going to be a very personal take. So I'm curious how the rest of the world feels. But I think the biggest part of content for me is the ability to share it, to go on Reddit, see what people are saying, talk to each other about it. So the way I see this heading, the AI native social media, right now users are generating it using AI. What does stop AI just responding to you? Because it's going to know you better than anyone else. So now you have no one on the other end. You have hyper-personalized, basically, drugs that are built for your dopamine receptors. that are...

8:04Yeah. And they can, you know, I mean, they can, people can build whatever content they want. I mean, the barrier is just so low now. So, I mean, I take your point. It's a little bit scary in some ways, but, you know, like you say, the social media angle to all this is really what is interesting. Sonya, thank you so much for coming on. We have so much more news to discuss in the weeks to come. We obviously have OpenAI's Dev Day coming up on Monday, and I think we are eagerly waiting to see what else they unveil. Thank you so much for coming on the show. That is Sonia Oja, a partner at Bain Capital Ventures.

8:38Okay, well, this week, a company co-founded by former energy secretary Rick Perry went public. It is called Fermi America, and it's building what could become the largest data center complex ever in Amarillo, Texas. Shares jumped in its debut yesterday, but the company is only a year old and has no revenue yet. Joining us now for more analysis is AI and finance reporter Miles Krupa. Miles, welcome back to the show. It is great to have you. Thanks, Akash. And I have to correct you off the bat really quickly. As a Texan, it's pronounced Amarillo. Can you tell that I've never been? I don't know. No, I have been to Texas.

9:15Yeah, I have no excuse. I certainly have never been to Amarillo. Well, it's an interesting place. Texas is a vast, vast state. Okay. You know, I got to be honest, Miles. This story was so crazy. That is not where I thought you were going to start here because the story itself has so many colorful angles to it. Let's just start by, did I pronounce Fermi right at least? Is that the name of the company? Yes, that's correct, Fermi. All right. Okay. So what is Fermi's prospective business model? What is it that they're trying to do here? Yeah. Fermi is basically in the business of procuring land and power for data centers.

9:58This is a really hot area right now because the big tech companies, they're building data centers as fast as they can. And the U.S. grid is very constrained. And so they're looking everywhere they can for land that has ready access to natural gas that could be good for nuclear development. So Fermi is basically trying to satisfy that demand. Okay. And so this is kind of like a real estate plan. I mean, we've heard about so many real estate companies getting into the data center space. Who is this company run by? Yeah, so it's run by Rick Perry, as you mentioned, as a co-founder and director. And then the CEO is a man named Toby Neugebauer.

10:41I hope I pronounced that right. It's a German-Polish name. And Toby Neugebauer is a longtime private equity executive. and he previously started a company called Glorify that was going to be a sort of banking app aimed at conservatives. It called itself Anti-Woke. It actually went bankrupt in 2023 after it failed to go public through a SPAC. So this is sort of his next business venture. Got it. Now, talk to me about the company. It is less than a year old. One of the things you mentioned, and this was from data from our friend Jay Ritter over at the University of Florida, He mentioned that it is the first REIT to go public since 2017, or at least that was according to his data.

11:27The first REIT to go public since 2017 that has no revenue, and yet shares popped yesterday. So what are we to make of this? Yeah, it's an interesting moment in the public markets for any company that is trying to build data centers. You can just see sort of the meme stock traders really latching on to this thesis. that because we're so power constrained, anybody with any sort of access to land and power is, it's almost like a biotech bet. It's like these companies might not have the goods quite yet, but there's potentially a lot of upside if they do have the goods eventually. So that's kind of an on-view here.

12:14Do they have any customers lined up at all yet? They have a letter of intent with one customer that's interested in being a tenant at a data center at the Amarillo site. But, you know, crucially, it's only a letter of intent, which means it's non-binding. So that customer could theoretically walk away at any time without many repercussions. Right. Well, these letters of intent are making a lot of noise these days in AI, that's for sure. You reported that the company is planning a facility called the Donald J. Trump Generating Plant. I assume this means there is obviously some history with the company and the Trump administration.

12:53Talk about that for us. Yeah, well, I mean, clearly, you know, Rick Perry has his government connections. You know, the company is working closely with the U.S. nuclear regulator on sort of a fast track way to get nuclear approved. You know, this is infamously something that takes many, many years and is sort of a very uncertain timeline. So, Fermi is trying to do this faster, and it seems that they may be leaning at least partially on flattery to get that done. Correct. And last question for you. I mean, you have covered debt for us here at The Information in ways that we haven't actually seen that much of in years past.

13:35Tell us about the debt angle of the story. They have to go out and raise this debt. Do you actually think they're going to be able to do that with the business model they have? Yeah. So, you know, the banks that they worked with on the IPO, they shared projections with potential investors showing they'll need to raise about$9 billion of debt over the next two years. The crucial thing there will be basically turning that customer LOI into an actual binding contract. You know, most lenders to data centers want to see a binding contract where they can project out the cash flows before they actually lend money against the projects.

14:10So until they, you know, secure at least one customer, I think it's going to be a really tough uphill battle for them. Great. Well, the beauty is the company is now public, so it means we can follow each and every one of their milestones out in the stock market. So thank you, Miles, for coming on. It's a very interesting story. And the combination of data centers and debt is obviously not one to take lightly in this climate. That is Miles Krupa, who covers AI and finance here at The Information. Okay, well, every week we hear of new companies, new AI companies hitting big revenue milestones. Companies can grow from no revenue to$100 million in annual recurring revenue seemingly overnight.

14:52But that doesn't always mean that those companies have enduring businesses. And to talk about that, I want to bring on Jammin Ball, a partner at Altimeter Capital who has done some great writing on exactly this topic. Jammin, welcome back to the show. It's great to see you again. Thanks for having me. Excited to chat today. Yeah, let's nerd out about all these metrics that are going in all different directions these days. Look, we had all these companies growing quickly, and yet the gist that I got from the post that you wrote was that you are kind of nervous right now. You mentioned it. You said you're a little bit anxious looking at some of these companies.

15:30I mean, things are so fun right now. Why is Jammin' Ball nervous right now? Yeah, look, I mean, I think it's important to call out first, right, that what we are seeing is some of the most historic growth trajectories that we've ever seen in any private software company. And it's for a few reasons. It's because with a lot of these AI products, whether it's an AI marketing agent, an AI sales agent, an AI finance agent, they're able to demonstrate ROI very quickly. You have a buyer who is very interested in applying AI and efficiencies to their own business, and a world that's just moving so fast that you can't risk to wait and stay on the sidelines.

16:08You have to go procure these things. Otherwise, your competition will, and you'll be left behind. Wrap on top of that, these are greenfield markets. A lot of these AI solutions aren't replacing anything. They're net new. Sometimes they may be augmenting or replacing some form of labor, but it's greenfield. Now, with that being said, the topic of the post I wrote last week was called Easy Come, Easy Go. I've maybe written a prior iteration of that, which is kind of ARR versus ERR, kind of the ARR being annual recurring revenue, very sticky. ERR, experimental runway revenue. But yes, I think there's – and we can chat all about it.

16:48There's definitely pros and cons to these historic ramps and growth rates. Well, so one of the things you basically talked about is basically some of the reasons why this initial surge in revenue for these companies may not be as sticky as other revenue that may have taken maybe a couple of years to generate. And so talk about some of the reasons here. You talked about implementation time, sales cycles. Yeah. What are some of the biggest concerns that you have or flags that should be raised? Yeah, look, I would say it's a beauty and a curse. And there's things that are very good, but may also have a negative, right?

17:25And there's really kind of like three I came up with, right? One was the implementation cycles. Well, actually, let me even start before that. The sales cycles are extremely short for a lot of these AI product purchases. And that's generally because of the few things I listed earlier. The time to value is very clear and obvious. You have a buyer who's very motivated, but in general, you have very fast sales cycles. On top of that, you have very fast implementation cycles. A lot of these products can be PLG'd. And you have a champion that is making a very fast decision. Again, I think it's very easy to hear me say those things and say, well, that sounds amazing.

18:03Isn't every company trying to lower their sales cycles? Isn't every company trying to lower the time it takes to go live on the implementation? Why are those bad things? And here's what I would say. They're not necessarily bad, but I think they can be taken for granted. When you have a very fast sales cycle, well, what does that probably mean? It probably means that the buyer didn't evaluate you against four or five other solutions. They moved really quick. It might mean that your champion really wasn't fully bought in. They made an emotional decision because their boss was telling them we have to move fast.

18:38And they weren't as bought into the solution. You were just a solution to an immediate hair-on-fire problem they had right now. And you mentioned that. In some cases, you mentioned they could have been the only vendor that was evaluated, which is crazy to me. That's not the way software typically worked, right? It hasn't, but I would say there's never been the pressure from higher-ups, from execs to move quickly and adopt AI. And when you have kind of like the pressure coming down from the top, when you have an employee base who's excited to adopt a lot of these solutions and tools and automate away a lot of the mundane parts of their day to day, you do have this perfect storm of just moving really quickly.

19:22And then there's also generally very fast implementation cycles. At the end of the day, you can't go zero to 100 in a year or two years without some level of fast implementation cycles. if it took a long time, you just probably couldn't grow that quickly. But that also probably means you're not necessarily, well, why are implementation cycles long? Because you have to integrate with a lot of solutions. You have to integrate with a lot of other workflows. You have to get different roles and access controls and all of that stuff, which again, people want to bring that time down, but the beauty of longer implementations, more integrations, kind of more workflows is, it's what I call like the Salesforce effect or like the Atlassian effect.

20:01Like these are products that you don't really find a lot of people saying, I love Salesforce, like I love Atlassian, but they can't rip them out because they just have their tentacles in a hundred different workflows and removing one solution actually means rewriting the hundred workflows that touched it. And so that is the beauty of a long implementation cycle is you're gonna really push your tentacles into workflows in all different other parts of the org. And so absent that, it might be easier to rip out and replace. And so again, it's kind of like the beauty and the curse here of shorter implementation cycles and shorter sales cycles.

20:37Let me ask you about what the right metric is here, because you talked about ARR versus what you sort of call experimental revenue run rate. I said, experimental. Yeah. Experimental run rate revenue. Run rate revenue. So, I mean, you've got the revenue metrics, right? The other metric we've heard so much about is net dollar retention, of course, which is a way to sort of measure this over time and see how many customers are actually sticking around or even upgrading in some cases. The thought that I've had, and I'm asking you this, you know, one of your thoughts here is, I mean, are metrics sort of becoming less important for many of these companies, even if they have revenue, you know, are the qualitative factors really around, hey, who is the customer?

21:21You know, what is the profile of the customer? Is that really what you have to focus on in this climate is leaning on your qualitative judgment of who these companies are selling to. I think, yeah, look, I think that's certainly part of it. I think the other metrics that I will lean a lot more on these days are kind of like gross retention and looking at retention cohorts, because that does tell a good story on how these customers are trending. Because again, I'm not saying that all of these customers are kind of like easy come, easy go. There will certainly be a number of companies that, you know, we see grow to be multi-billion dollar, tens of billion dollar, large public companies that started off maybe with a profile that could have let easy come, easy go, but just continued their liftoff.

22:04The execution rate we're seeing of a lot of these companies is faster than ever. But I'll look at other metrics. But yes, I think it's almost like counterparty risk. It's customer risk. It's, hey, let's go talk to that customer. And what you might hear sometimes is, hey, we bought this product because we wanted to build this feature we don't know what the exact roi of that feature will be but we have a hypothesis we think it will drive higher conversions we think it will drive a higher asp we think it will lead to better customer satisfaction right um but at the end of the day if it doesn't maybe they'll just turn it off and stop using it but they can't afford not to try it um and so they're trying a lot they're experimenting a lot and again a lot of those experiments will be shut off but also a lot of those experiments will be double, triple, quadruple down on.

22:54And we'll see kind of a new crop of historic companies built. Well, I mean, that was a big point that came up in our AI Agenda Live conference this week was, you know, companies saying, hey, the pilots are what they are. Don't stop or start spending based on the pilot. Just, you know, keep chugging along and seeing how they go. Jamin, thank you so much for coming on the show and explaining it to us. It's a great newsletter. Again, he is the author of the Clouded Judgment newsletter. That is Jemen Ball from Altimeter Capital. Okay, well, speaking of our AI Agenda Live conference, I am still on a high from the event earlier this week in New York City.

23:28We had a number of great speakers, including Sarah Guo from Conviction and executives from Reflection AI, Weka, and OneX. We've got a quick video of some of the highlights. Let's take a look. It's AI. It's like the topic du jour. Everyone get excited. I'm going to name you Robert the Robot. What? I think we will all have our own personal robot in the next decade. There will be at least 8 billion humanoid on the planet. And they'll be our friend. They'll be our companion. They'll be there throughout our lives, helping us with everything we do. And I think there's a lot of magic to that. NVIDIA is a great partner.

24:09I've known Jensen as another joke goes like$4 trillion ago. when they were a normal company and he was a normal person. For us as investors, we look at sort of what is the unique spike of the founder? What's the unique insight? What is the unique data? What is the unique data flywheel? What is the actual workflow that needs to be built? Because all of these things, you know, I need to find a better phrase for this, but I call it elbow grease. You got to put in the elbow grease and the grit to actually make some of these amazing businesses. Anything that can be vibe coded in a weekend is not a valuable thing to invest in.

24:43We want to have AI mathematicians, super intelligent reasoners to collaborate with mathematicians, apply scientists to figure out how to scale formal reasoning in existing markets and then to expand new markets and new use cases. One of the things that's undervalued in AI is like this is a relatively small industry and set of researchers and small dollars put toward it like as recently as 2022. too. And so I do think it is harder to make progress from here, but I absolutely believe that we're going to see more capability. Thank you, everyone, for joining us today. It is a pleasure to see you all.

25:22Have a good night. That was the highlights from our AI Agenda Live conference here in New York City earlier this week. Okay. Well, more and more, we are seeing big tech companies gobbling up smaller startups. Just today, we saw HR software company Gusto buy Arrival for$600 million. We heard just yesterday from Bursell about how it has plans to continue its acquisition streak. And then we also have big public companies like Atlassian that has been buying well-known companies like the Browser Company. I want to bring on Francis Aquila, one of Sullivan & Cromwell's top M &A lawyers, to talk about what he's seeing in the current deal-making environment.

26:02Francis, it's so good to see you. Welcome to TITV. Hi, how are you? Thanks for having me. So I'm really excited to chat with you about what you're seeing in the market. Look, from our end, it feels like M &A is up. And I wonder from your end, is that what you're seeing in the data? I mean, talk to us about what you're seeing in your practice. What we're seeing is, well, first of all, I'll go back to the beginning of the year. I think everyone in the M &A world anticipated a tremendous amount of M &A activity. We anticipated that tech, particularly the AI space, but also non-AI, in effect, driven by the changes that AI is bringing about, creating a lot of M &A activity.