Sam Altman’s Warning About Google, Musk’s $1B+ Lawsuit Against OpenAI, AI in Travel | Nov 21, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Pasricha. It is Friday, November 21st. We have a busy show ahead of us. First up, we've got the inside scoop on a warning OpenAI CEO Sam Altman shared with employees in a recent memo. We've then got Redpoint Ventures coming on the show to talk about assessing early stage bets in 2025. And we've got the CEO of Booking Holdings coming on to talk about how AI is reshaping the travel economy. We're also bringing you the second installment of The Editor's Cut, where we take you inside the information's edit meetings to unpack the biggest questions inside the tech industry.

0:53Today, we're focusing on the fierce rivalry among the big frontier model companies. And finally, we're closing out the week with a status check on the Elon Musk OpenAI lawsuit. And given recent developments, how much or how little is at stake now for both parties? It is a big show. And so let's get right on into things. Last night, my colleagues Stephanie Palazzolo and Aaron Wu published a great piece about a cautionary note that OpenAI CEO Sam Altman sent his employees. It is a rare look into the inner workings of the company, and I want to bring on Stephanie and Aaron to tell us more about it.

1:29Stephanie and Aaron, welcome back to the show. It's great to have you here. Hey, great to have you. So let's talk about this message, Steph. What was the note that Sam Altman sent to his employees? Yeah, so as Aaron and I reported yesterday, we were able to get a hold of an internal memo that Sam Altman sent OpenAI employees late last month. So this is, again, you know, weeks before Gemini 3 even was released. So they were clearly already hearing rumors then that the model was going to be very good. And, you know, in the memo, Sam is honestly pretty upfront with employees. And he does admit that the Gemini model will be very good.

2:07And he tells staff to expect temporary economic headwinds, as well as kind of just generally rough vibes for a while after the release will come out. And he does also admit that, you know, the Gemini team has done really good work, especially on a certain part of the model training process called pre-training. But he does end on a somewhat positive note saying that he still expects OpenAI to come out on top. And there are lots of ambitious bets that the company is taking that he thinks will pan out for it in the end. I'm going to come back to pre-training in a couple minutes here. But Steph, this is kind of a broader story here that we've been covering at the Information, which is that OpenAI had this lead ahead of all these other Frontier model companies.

2:50And it's not just Google. I mean, other companies are catching up now, right? Yeah, no, it's been a very crazy past couple of years. But definitely, as we get further away from the ChatGPT release, we see that that gap between OpenAI and other players like Anthropoc and Google has definitely narrowed. And I think from what we hear from developers and founders, it does feel like every couple of months or so, maybe one lab will come out with a model that leapfrogs the others. But essentially everyone is around the same place. I guess where things do differ, obviously, is in the actual usage of the products that these companies are coming out with.

3:32So even though Google and OpenAI's models do seem very like neck and neck these days. We do see from usage that, you know, OpenAI's ChatTBT has 800 million weekly active users while the Gemini app has 650 million monthly active users. So both a lower number of users, but also these users are not going to the app as often it seems. So Erin, how did Google do it? Because it feels like 12 to 18 months ago, this was not the conversation we were having around Google. Absolutely. So, I mean, this is a great question. This is the multi-billion dollar question, like a multi-hundred billion dollar question.

4:14Unfortunately, I think we don't totally know. Like when looking at what Google said, looking at what one of Gemini's tech leads, Oriol Vignailes, said when this model was released, he said, you know, the secret, the way we did this, we just made a ton of progress on pre-training and post-training. And it's like, okay, that's sort of the whole thing. And so Google's doing Welcome to AI. There's pre-training, post-training, it's inference. It's the whole thing. Congratulations. We made a better model. So Google's doing a really good job on a lot of different levels of the AI training stack right now.

4:46And I'm talking to a lot of Google researchers and a lot of people are saying, look, it's a question of like, we have the right people, like we're going after the right ideas. We've written some about how specific people have had various impacts. A couple of weeks ago, my colleague Sylvia and I profiled Noam Shazir, who's a very controversial figure in a lot of ways, but people are also telling us, look, no, he came in and he did this improvement to the pre-trading stack that justified the whole $3 billion, the$2.7 billion price of the character deal that brought him in. So I think there's a lot of different pieces that come into this, but what one of my most skeptical AI sources said was like, look, this feels like it vindicates the idea that if a big enough lab that is going after the right things, like, organizes its people in the right ways, like, there doesn't seem like we've hit a limit yet to, like, what AI can do.

5:43So, Steph, this kind of brings us back to a very core question, which actually was, it was one of the first comments on the story was, see, the models are just being commoditized. It actually doesn't really matter whether Gemini 3 is better than, you know, GBT5, whatever. I mean, they're coming out so fast. We're going to stop talking about these models eventually at some point. Is that the idea? I mean, again, this is the big question. And, you know, this is a question that many investors are thinking of whenever they decide to pour billions of dollars into these labs. Hundreds of billions. Yes, hundreds of billions at this point.

6:23It's very crazy. I think it's fair to say that at this point, it seems like all the labs are kind of at around the same place. And so I think because, you know, researchers are jumping from lab to lab all the time. It's very difficult to keep some sort of technique, you know, secret at a lab for any amount of time. And so basically, I think all the researchers at the different labs, know what the other ones are doing. No really secret sauce there. I do think, though, that as Aaron was kind of talking about, there are really amazing researchers at these labs, and they're still making slightly different bets on different types of techniques that they want to focus on.

7:06And so, for instance, this is what has helped Anthropic continue to keep its model to be so good at coding, even while OpenAI, Google, everyone else also wants to be really good at coding as well. And so, yes, like they're all generally at about the same place, but each one can still have the opportunity to have a little edge in certain niches. Aaron, last question for you here. Let's come back to sort of some of the fundamental advantages that Google has compared to OpenAI, not the least of which is the fact that Google generates not just cash, but a whole lot of it. And OpenAI has to raise, you know, almost the same amount of cash that Google produces in a single year in some cases.

7:49Can you talk a little bit about the, you know, how your sources are thinking about just how quickly Google can pull ahead maybe a year or two from now just based on how the business is structured? Yeah. So, I mean, like you said, Google's big advantage is that Google has so much money. Like Google does not need to raise money because Google's trailing free cash flow over the past 12 months is around 75 billion, which is more money than OpenAI has raised in its entire existence. And obviously, this has to go to a business that's supporting a lot more things than OpenAI. But the idea is Google has a sustainable business, even if AI is not directly, even if they're still able to pour a ton of money into training AI models.

8:36And so the idea is essentially Google has a lot of advantages. They have a huge scale. They have all of these products with over 2 billion users already. And so I think that Google is in a really good place right now to be able to start figuring on its product strategy, to be able to start putting Gemini into all of these things. And once it's able to do that, it's in this much more sustainable place than OpenAI is where OpenAI needs to keep raising money. OpenAI has predicted that it's going to burn more than$100 billion in the next few years. And so that's all money that has to come in. And Google is just not in that place.

9:16Right. Great. Well, Steph and Erin, it was some great reporting. I want to thank you for coming on and we'll have you back on the show again soon. Thanks. Okay. Well, our next guest is Managing Director at Redpoint Ventures. She was previously the CEO of GitHub and she has made a ton of cool investments in sectors like cybersecurity, developer tools, and yes, AI. She also previously co-founded a software company of her own that she sold to VMware. Erica Brescia, welcome to TITV. It's great to have you on. Thanks so much. Great to be here. I was COO of GitHub, by the way. COO? What did I say? I think you said CEO, but it happens all the time.

9:54I've never cracked with anyone, and I'm like, I should probably make it clear. No, no, no. Thank you for that. Look, I had COO on the teleprompter in front of me, actually. So that's just Friday brain from my end. So thank you for correcting me on that. Look, Eric, I want to talk to you about a number of different things. I want to talk to you about your approach to venture investing, the sectors that you cover. But very quickly, look, I do want to pick up on the discussion we just had in the previous segment, which was around the lead that OpenAI had in frontier models and how that's narrowing a bit as the other model companies get better and better.

10:32And the broader question I just want to ask you is this question of commoditization in models. You know, we are talking a lot about them now. Will we be talking a lot about this three years from now? And if not, how do you think the structure of all these big AI companies changes in the long run? It's a great question, and I think the question, my view is, I don't know if it's three years or five years because we're still so early in this race. But I think you're right. I don't think we're going to be talking about the models as much. I do think they're largely going to be commoditized. But I think they're going to be commoditized in the way the cloud vendors have been commoditized, right?

11:12You know, there are still incredible businesses that are printing cash, whose margins have improved over time, who have built all kinds of different services on top of the base substrate of the platform. And they all have a different feel and different ways that they spike, right? People go to Azure for different things than AWS, than Oracle, et cetera. And that's been my thesis actually since about, you know, day zero of this. It's the best mental model that I've come up with for how to think about how AI and the model vendors in particular evolve over time. Right. And so this shift to applications, you know, this is something everyone's talking about.

11:54I mean, that seems to, you know, OpenAI just hired Fiji Simo as the CEO of applications. Look, we have coding as one application. We have agents. You know, where are you looking right now in your investing with respect to what applications you think are really going to take off here? Sure. Well, as a firm, we invest about half in infra and half in apps. I want to be clear. I think there's a ton of opportunities still at the infrastructure layer for investing. You know, if you look at a company like Snowflake, came along six years after AWS was founded. And I think it's important for us all to remember that.

12:31A lot of the key infrastructure for building applications or getting value out of these models has yet to be developed. A lot of the times it comes from people trying to build something and put it into production in a big company, having to build a bunch of tooling around it, and then spinning it out. So I'm still spending a good amount of my time in cyber and infra. We also have been very active, along with all of our peer firms, at the application layer. And there's a couple key themes there. One of them is we're really looking at defensibility in these businesses. And the way that we think about that is what workflows are you taking over and across what systems are you integrating?

13:16I think just being an API call gets really challenging. And I know there was a whole, you know, wave of discussion around API wrappers. There's nothing wrong with building on top of these models. But we think the defensibility comes when you're really integrating with different systems across a company that make it, you know, much harder to rip something out. I think an interesting observation on that front is, you know, if you're one of the first companies to get in, take one of our portfolio companies, this company Liberate in the insurance space, you know, they can deliver like a 10x improvement over existing processes because they're completely replacing human labor and automating workflows across a bunch of different systems.

14:01Now, if another AI company comes in and tries to sell to their customers, it's going to be really hard for them to deliver a 10x improvement over that. It starts getting to be incremental improvements. Let me ask you about the cyberspace, which is an area that you focus on a lot with your own investing. Sure. What is the story here in cybersecurity right now in this moment? We saw the Anthropic report about the threats that AI could pose to cybersecurity. Of course, there are people saying that, yeah, I mean, they can use it to attack. We can also use it to defend in both ways. But, you know, we've had cybersecurity CEOs on the show that have been open about the fact that said, look, the threat is, you know, in some ways actually outweighs the benefits as of right now.

14:51Do you see this as a concern for you? Absolutely. It's something we've been thinking about for quite some time. You know, there's this escalation war. Every time we create something great and, you know, great technology makes its way into open source, too, there are bad people out there that will figure out how to do nefarious things with it. And so a big part of our investment thesis in cyber has been, look, the attack landscape is getting ever more violent and active. and we need to help companies and good actors with tools that allow them to do everything that they can to beef up their own defenses.

15:33And that goes from, you know, doing pen tests once a year to automated AI-driven pen testing. It goes to, you know, enterprise tooling to help provide gateways and control access and provide better protections against, you know, ever-increasing threats out there. It includes investing in human risk platforms that help companies either teach employees or preventing them from falling for phishing attacks and things like that. So we've been very active. Let me ask you one last question before you go. I was reading up on your bio and before you were at GitHub, you started your own company and you actually went through Y Combinator but more than 10 years ago now at this point.

16:19And I just wanted to ask you how you think about accelerators in 2025 broadly. Who should do an accelerator? Who shouldn't do an accelerator? We have a lot of people watching the show that are thinking about starting their own companies. Has the structure or the value prop or the methodology around thinking about that changed at all? Yeah, it's a great question. And to all those thinking about starting a company, No time like the president or the president. It's an incredible time in history to do that. I did have a very different experience with YC. My batch was 42 companies. They are several hundred now.

16:56I think what you really get is the companionship, the connections, a kind of brain trust that you can bounce ideas off of. And honestly, there's a lot of very healthy competition. You know, somebody is going faster than you, shipping faster, growing faster. It's a really powerful motivating factor. And I think community that you get to be a part of and see what's working and what's not across a lot of, you know, very highly capable, highly driven, special people. So let me ask you, who shouldn't do an accelerator? Maybe that's a better question. Yeah, you know, look, if you have spent a lot of time in very high growth, successful companies, and you feel like you already have a peer network that is very strong of people who are building companies, you can share advice with your learnings commiserate with, you might not need to give up the equity to an accelerator.

17:54it was useful for me. I didn't go to Stanford. My parents weren't in tech. We started my first company outside of the US. I had a Spanish co-founder. And so I really felt like a Silicon Valley outsider. And it was really helpful for me to build my network and just get in with a group of people I could learn from. If you don't need that, you probably don't need an accelerator. Great. Well, Erica, thank you for coming on the show. We really appreciate it. And it's a great discussion all around. We'll have you back again soon. Great. Thank you. Okay. Booking Holdings is one of the biggest travel tech companies around.

18:30It is not one that we talk about a lot on this show, but it is the parent company of Booking.com, Priceline, Kayak, and OpenTable. And for context, even if you combined Airbnb and Expedia, that would still only be about two-thirds the size of Booking Holdings market cap. And yet, many have raised questions about the threat that AI has to the travel industry's businesses. Google unveiled new AI-powered tools this week, causing many travel stocks to stumble. Of course, Booking has been embarking on its own AI strategy with a big partnership with OpenAI. Joining me now to discuss all of this is Glenn Fogel, CEO of Booking Holdings.

19:05Glenn, welcome to TI-TV. It's great to have you here. Well, I'm so happy to be here. Well, I'm excited for this conversation. Look, where I want to start is the news that happened earlier this week. Google came out with all these cool AI tools. We mentioned it up top. We saw some of the stocks take a bit of a dip. How are you thinking about those AI tools as a competitive threat to your business? Well, let's start with just AI in general and how incredibly exciting it is for the travel industry and how pleased we are to be part of it and what we're continuing to build out in AI, which is fantastic.

19:38Now, working with our partners like Google has been really great, and we envision a great partnership going forward in the AI area. I loved yesterday when they came out with a little bit of a news piece saying, listen, we're not planning to be an OTA, which I love for them to qualify because I think there may be some misunderstanding. And OTA for the people who aren't as versed in the travel industry. OTA is? Online travel agent. Oh, there you go. Okay. Pretty simple. And that was a little confusion, I think, at the beginning of the week. No, this isn't news at all. We've all been working. how can we use AI, say all of us, I mean all the people in the travel industry, and also the people that are frontier players who are creating these wonderful new technologies, whether it be Google or OpenAI or any of them.

20:26And we have been working with everybody. You started off, you mentioned about our relationship with OpenAI and what we're doing with ChatTBT, which is great, the things we've done with them. But we're doing a lot of working together with Google, of course, and also others. look for with Amazon and what they're doing with Alexa. Every single player in the space, us included, of course, we're all working with the different players who are doing the technology. So there's no news here at all, though it does seem if you look at the market in terms of stock valuation, you would think there was something new happening.

21:00Right, right. Well, so I do want to get sort of inside your head a little bit. The OpenAI deal was really exciting for us here in the newsroom because we were really curious to see which companies would partner with OpenAI and all their efforts to sort of bring everything into their platform. I'm curious, when you were thinking about whether or not to do that partnership, there were a lot of arguments for, was there anyone in the boardroom that raised arguments against doing the deal? What were those arguments, if there were some? Well, first of all, it doesn't go to the board level. Well, I'm not talking about the board, I'm talking about the conference room, fine.

21:38So look, we always want to go where the customers are. And if customers want to begin their travel search, their travel investigation, what they may want to do in travel, we want to be there for them to help make sure they get what they need, make sure we continue to provide the service we've done over the last 25 years I've been at this company. And whether it was they wanted to start off at just the beginning with the old website with a blue link at Google, or now wanting to go to a large language model to begin their investigation. That, to us, is just a way that customers are starting. What we want to make sure we're there, we're there so they can get what really is necessary, which is a great experience in producing a travel thing they want that is so much better than it used to be before there was technology.

22:27Right. Talk to me about the results that you've seen from that partnership on your business. My God, it just started. Come back to me in a year. I'll tell you how it's going. But I mean, it hasn't adversely affect traffic. You mentioned this on the earnings call. Traffic for your own search product has still been growing. Tremendously so. And over the last, you know, whatever, since Google started, regular search continues to grow. It's been a wonderful relationship with Google. I imagine we'll have a wonderful relationship with everybody who's coming in with new ways for travelers to investigate.

23:02their needs. Right. I want to talk about a side of your business we don't talk about a lot, which is the B2B side of your business. For people who aren't following the company as much, who are you selling to? What is the product? And how is that side of the business faring? Well, the B2B business you're afraid to is when we partner with somebody who has their own customers who want to have travel services, an obvious one would be a bank. So let's take an example, Citibank. We provide for Citibank customers who want to get certain sort of benefits, being a Citibank consumer, whether it be just an account there or they have perhaps a credit card and they are able to get their travel services fulfilled.

23:46Well, if Citi needs somebody to actually do that, that would be us. That's an example. Or if you're an airline and you sell flights on your own, but you need somebody to provide the hotel benefit, We would do that for a big airline. That's the nature of that type of B2B business. And how big is that business now? It's big. We're not as big as the leader in the world, but it's growing, but we don't break it out. Right. And, you know, going back to AI, I am curious, you have more than 20 ,000 employees now at the company. You're using all types of software, I'm sure, to manage employees, help them sort of carry out their day-to-day activities.

24:24Have you used AI at all internally to cut costs, make things more efficient? How are you thinking about that? Well, we talked about this on our call. In fact, many calls. You go back several quarters and we've talked about how much AI is improving the efficiency of the company. So we do have about 25 ,000 employees. One of the really important things is providing great customer service. And that's an area where we see incredible benefits using Gen.AI to make so many ways to improve that, what can be for many people, a very frustrating experience. So just right away, nothing's more annoying in customer service if your customer is waiting for somebody to pick up the phone.

25:05Well, with Gen.AI, you don't have anybody waiting to pick up the phone as being picked up by our technology. or just want to get that simple answer to a problem or a question you have automatically without even having to deal with a call, we provide those services too. It's one of those areas where we see incredible benefits that's only going to improve. Open Table is one of our companies. So sometimes you call a restaurant for a reservation. You think you're talking to a host to get their reservation. That's not a human being. That's AI. or if the restaurant needs some questions about the service that we provide to them at OpenTable, the restaurant, when they make a phone call, is also not talking to a human being.

25:46It's to an AI agent. These things are going to just continue to improve, and whether it be coding, whether it be any type of service that we're providing, AI is going to continue to improve the efficiency. Right. You know, I am curious, you've been at the company for 25 years now, I think, and it's rare for anyone to stay at a company for that long. A lot of people watching this show, they have hopes to start their own businesses and you continue to climb the corporate ladder. What is the perspective from your end on why people should aim to climb the corporate ladder in 2025 versus starting their own business?

26:25Well, I think it's up to the individual. I am not the person to give any advice to anybody in career. I don't even give my kids advice in terms of they're out in the work world themselves. It's up to individuals and what kind of lifestyle you want to have, what kind of risks you want to take on. But what is the argument for climbing the corporate ladder in 2025? Climbing the corporate ladder just has to me some sort of, let's say a negative connotation. I believe continue to develop your abilities, your knowledge, providing a better way to do your work and provide to see the results of what you put your effort into is a great thing.

27:07Whether that be done at one company or you feel that company is longer providing you what you need to continue to advance, so you go to another company, that's fine. I myself have been very satisfied. You know, I joined this company. You know, at a point in my first year, our stock price was$1 a share. Now, that was pre-reverse split that we had to do because we were below$1. So let's call after the reverse split$6 a share. In the summer, we got close to$6 ,000, right? Now, it's a little bit down right now from that high. But think about that. Look at that growth. And in every single part of my journey, I've had the opportunity to do different things, have different challenges, learn different things.