EA’s $55B Buyout, Snowflake’s AI Data Strategy, Humanoid Robots Learning Dexterity | Sep 29, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Pasricha. It is Monday, September 29th. We have got a great show planned for you today. We are talking about the Electronic Arts Takeover and also about Silver Lake, one of the biggest PE firms involved in the deal. We've also got Snowflake coming on to talk about their answer to the corporate data wars that have been sweeping through AI. We are then going to take a step back and look at the scale of power that NVIDIA has amassed with its investments, not just into OpenAI, but also so many other pockets of the AI ecosystem. And last but not least, we are bringing on the man behind the popular Roomba robot to talk about some reality checks that he thinks everyone should keep in mind as it relates to humanoid robotics.

0:57We've got a busy show, so let's get right on into things. Electronic Arts, the video game company behind several popular franchises, including the Madden and NHL games, is going private in a$55 billion takeover. It is one of the biggest leverage buyouts that we've seen, and I want to bring on our Deputy Bureau Chief of Finance, Corey Weinberg, to help us break it all down, including some analysis on the firms that were involved. Corey, welcome back to the show. It's great to have you again. Good morning from Tinseltown, Akash. How's LA, man? This is the first time you're joining us from the west side over there.

1:35Yeah, no, I'm happy to be back on the best coast and always great to be with you. okay well let's talk about the the news today so ea is going private for 55 billion dollars what did you make of the deal i mean we haven't really seen uh sort of a huge kind of like uh ground shaking private equity deal in in a while i mean interest rates have been have been high the PE industry has been nursing its wounds to a large extent. A lot of times when we come on, when we talk about these big private equity deals in tech, it's a pure software play. It's like Silver Lake had bought the software from Qualtrics before.

2:25To be able to talk about a big consumer deal is kind of fun. And this one obviously had some interesting wrinkles to it that we can get into. Well, talk about those wrinkles. I mean, the business itself, EA, video game industry, I feel like has been changing. It's sort of been going through a similar shift that we saw with movies and entertainment, right? I mean, you went from buying single games to now the subscriptions business. The company itself, revenue is flatlined. Cash flow has also been pretty flat. What do you make of the business itself? Yeah, look, it's kind of a classic tale of a business model in transition that, you know, folks in the private equity world would say is more manageable to undertake when you're in private hands than, you know, sort of facing public investors quarter after quarter.

3:19um so yeah i mean you've you've had a ton of sort of business model shifts you know specifically around subscriptions and you know sort of uh platforms like like roblox you know sort of being able to take a lot of the youth uh sort of share um you know you had a lot of big moments like this is it's a hits game you know gaming is all the space business right you know you gotta you got to make sure that next Madden title lands and things like that. So, you know, what we've seen so far early in sort of this deals commentary is a lot of talk around how could the sponsors use AI to kind of lower costs or to sort of not just lower costs, but also like boost like the amount of content that can be produced.

4:08And what about the investors? i mean what did you make of the group that came together to actually put put up this cash oh i mean like the thing that stands out immediately is the the yeah sort of the the importance of sovereign money this deal the you know the like saudi arabia coming in they already owned i think close to 10 of this business large gaming culture in the country they had made some other gaming related investments in the past but you know sort of seeing sovereigns which which as a reminder They obviously have gushers of cash from oil money. They don't have to go out and raise a fund.

4:53Oh, I think we just lost you for a minute there. Just rewind 10 seconds for me. We just lost your audio. Rewind 10 seconds. Yeah, I mean, the existence of sovereign cash is an incredibly important component of this deal. The Saudis are putting in, you know, sort of obviously tens of billions of dollars to finance this. And they brought along one of their recent best friends and Jared Kushner, whose private equity firm Affinity Partners is also a part of this deal. PIF, the Saudis, is the anchor LP to Kushner's fund. And obviously, you could read between the lines that having Jared Kushner, the president's son-in-law, involved in the deal means it could have a much higher likelihood of getting through the relevant regulatory authorities, particularly when there is foreign cash in a deal.

5:54What about Silver Lake? uh silver lake silver lake remains to in to my money the most interesting uh private equity firm in in tech uh they are often around some of the the biggest hairiest deals um we of course saw back to back weeks where silver lake was in the headlines they are obviously a part of the new Invest Your Consortium in the US version of TikTok. And it's a firm that has come out and on the record said, you know, we're going to be taking big swings. They raised about a$20 billion fund in 2024. And they said, look, we're done with small growth equity investments. We're done with sort of smaller, you know, sort of non-needle moving deals.

6:49We're going to do the big ones. And they already own Endeavor, which is the big talent agency, WME, as well as the WWE. And so they're often around big media, sports, sort of consumer deals, not just sort of the software deals that you see a lot of their tech private equity brethren doing. Right. I mean, it was kind of interesting to me to see their involvement in the TikTok deal, because like you said, I mean, private equity, you know, I don't know that I think of TikTok as like a, you know, a private equity centric investment. And yet here they are. Very quickly, Corey, before we let you go, I mean, you know, IPOs have been ticking up.

7:35It's a space that you follow very closely from the people you're talking to. Is this, are we expecting a lot more in the next two weeks or what are you watching for there? Yeah. I mean, I think we're definitely, we're entering a phase of the IPO market that's more normalized, where every single... It used to be for the last two years, every single IPO that came out, you watched with bated breath of, will they or won't they be able to get out? I think there's generally a sense with just how well the market has done, how well all the recent IPOs have performed, that it's generally a proposition that if you have a good company, that it will be greeted favorably from investors.

8:23And the fun part about that is we're going to see, I'm sure, a whole bunch of very speculative, not good companies also try to make that same pitch. And that's where it really gets interesting. So I think that'll be a lot of fun. In terms of the venture-backed tech software world. You talked about Nivan on the other week. That's going to be a classic one to watch and should be an interesting bellwether for other venture-backed software firms. Great. Well, Corey, thanks for being here. Have fun drinking the great weather over in LA. We miss you in the New York office, but I know that you're having a lot of fun meeting a lot of people out there.

9:09So that is Corey Weinberg, our Deputy Bureau Chief of Finance here at The Information. Okay, well, we have written a lot about the corporate data wars at The Information. As a reminder, this story has really surrounded the trend that big tech companies like Salesforce and Atlassian have basically been putting up walls of sorts to prevent AI companies from accessing data that they need to run their services. Last week, Snowflake joined a number of other tech companies and putting together a consortium that has agreed to let these AI apps and chatbots pull data from their systems. I want to bring on Josh Klar, a product management director at Snowflake, to tell us more about the initiative.

9:49Josh, welcome to TITV. It's great to have you. Hey, Josh. Thanks for having me. So tell me about this consortium. What exactly is the group aiming to do here? Well, it all started with an investment that we've been making at Snowflake around how do we unlock business semantics for AI-powered analytics? And what we found out is that these semantics, the definition of these business models, key metrics, the way companies look at their business, exists in a bunch of different areas. It exists in business intelligence tools and data pipelines in Snowflake. Do me a favor. Bring it down just one. So it's just semantics, data pipeline, like, you know, grade 7, 10.

10:29Grade 7. Yeah, yeah, yeah. Let's start there, baby. The language that customers use, our customers use, and businesses use to define their key business metrics like sales, net profit margin, return on ad spend. These business definitions exist in a bunch of different locations. And if customers want to be able to have AI-powered analytics, they want to be able to talk to their data. Instead of writing complex queries, they want to go to a chat GPT-like interface and say, hey, tell me what happened with my marketing spend and tell me why it happened. In order for that to work, you need to provide these LLMs, these AI powered experiences data that exists across a bunch of different systems, not just Snowflake system, but your CRM system, your upstream data systems.

11:13And so this is a problem that we've written about is that some companies have put up sort of some of these walls, say, you know, we're not going to let some of these LLMs access the data. This consortium, the way I understand it, is Snowflake and this group of companies way of saying, hey, you know, open access. There are no walls here. How are you actually, like, when a consortium comes together, is it just like a promise? Like, how does it work? It is a promise, but it's also an effort. There's a working group that we have created that is looking at the way that these definitions exist in all of our different systems.

11:48And we are developing a standard that makes it really easy to pass this information back and forth between parties so that an AI-powered experience in Snowflake or some other system can easily access data across all of those different interfaces. Right. Now, tell me about how the group came together. Is Snowflake going out and reaching out to these companies and sort of quarterbacking it? Are you getting inbound requests from companies to be part of this? I'd say it started with us and a few of our partners like Tableau and DBT. We kind of heard the same message. And in my discussions with them, we identified, hey, this is something we really should be doing.

12:24So we started this process of developing the open source consortium. We did a bunch of recruiting and lined up a bunch of additional partners. Since we launched last week, we've had something like 200 additional partners reaching out to us to try to be involved. So we're really, we've been quarterbacking it, but we really want this to be kind of an industry-led effort. And you mentioned Tableau. You know, I am curious kind of about Salesforce's involvement here because Salesforce is a company that, Candid, that we have written about at the International is one of those companies putting up walls.

12:57What did you make of their involvement? I think for us, but also my experience with working with Salesforce and Tableau, it's been about customer success. How do we unlock the success of what our customers are trying to do? And what Tableau is trying to do with their semantics is make sure that customers can access this data and have AI-powered experiences regardless of where the data sits. And so, at least from my perspective, it was really oriented around how do we unlock what customers are trying to do. And any pushback from other software companies in the industry? No, I've actually, I've been really surprised, I'd say pleasantly surprised at the level of support and interest.

13:36it seems like we've kind of touched a nerve that everybody is realizing we're not going to be successful with doing the things that we want to do, but also that our customers want to do without having this kind of openness. And so, I mean, you touched on the fact that you were pleasantly surprised. I just want to talk about, you know, how big a problem this was for not just Snowflake, but other companies. It sounds like this was a pretty significant issue for AI companies trying to go to their business. I think it's definitely a challenge. If you look at AI projects that kick off and then subsequently fail, there is an interesting MIT report recently that said 95 % of these things fail.

14:13It's often because they don't have the right prompts and the right context to be successful. And so this is, it's definitely a big problem. The analog that I've been thinking about is this idea of MCP servers. MCP servers really provide a way for AI bots or LLMs to talk to each other. And when you have this, you can actually unlock the reasoning and the power. And so I think the same thing is happening with data access. If you keep it a wall garden, you're just not going to be successful. Great. Well, Josh, thank you so much for coming on the show and talking to us about the consortium. We really appreciate it.

14:47We look forward to having you on again. That is Josh Klar, a product management director at Snowflake. Okay. NVIDIA has been very newsy lately with its Intel and OpenAI investments. But as usual, there are multiple sides to all these stories and also a number of different perspectives on why the company is spending all this cash. I want to bring on Anissa Gardisi, our cloud and compute reporter, to talk a bit about what our colleagues found in a weekend feature that they wrote about NVIDIA's cash spending spree. Anissa, welcome back to the show. It's great to have you. Hey, Akash. How many times a week are you averaging on this show right now?

15:24We should make it five. I'm ready for that. You want to be a co-host? Is that what you're saying? No, no, no. Okay, so let's talk about the feature. And then I also want to talk about the piece that you published today. But, you know, this line from this feature that you wrote with our colleagues stood out to me. The line about NVIDIA Spending Spree was, one possible factor behind NVIDIA Spending Spree is fear. And it sort of got me thinking about whether this is an offensive or a defensive play on NVIDIA's part. Why do you think fear is involved here? We wrote that line because of conversations we've been having with people at NVIDIA.

16:02And even though NVIDIA is clearly the dominant company in AI right now, there is a fear inside the company that there will be a major breakthrough in AI or another industry that isn't reliant on NVIDIA's ecosystem. And so that is something the company thinks about constantly. How do we stay close to the companies at the cutting edge so that we can make sure we are closely tied with what they are doing and, you know, a partner that they need in order to achieve all of their breakthroughs. And so that's, I think, why we see some of the NVIDIA investments that we have seen in the past few years. They want to be at the cutting edge and they don't want to miss out on the next big frontier.

16:46So it's kind of interesting to think about all the threats, I guess, you know, coming for NVIDIA. We, of course, have the startup chip companies, which I think are, they're still quite small in the grand scheme of things, right? I mean, the main threats that I think you outlined are, number one, the threat from China. Number two is a threat from the hyperscalers. And number three is a threat from OpenAI making their own chips, possibly. I mean, we don't have time to go through all three. Which of those three do you think are the biggest threat right now? I think the one, I mean, I think it has to be that the China market is going to be reliant on other hardware.

17:23That would be a major market for NVIDIA that all of a sudden does not need its chips. But I think the thing that maybe is more top of mind for Jensen right now is OpenAI. And he addressed that last week with the up to$100 billion investment in the company. The concern there is that OpenAI could be reliant on its own chips in the future and not need NVIDIA's, but today all the signs are pointing that OpenAI is heavily reliant on NVIDIA. Right. And I was struck by the news this morning. We saw news that Huawei is increasing their output now of their own chips to sort of possibly fill that market gap because NVIDIA is still having trouble selling in China.

18:01And then we also have DeepSeek, which they're rolling out updates that came out with another update today. And so, I mean, for me, I feel like the threat from these companies that are moving quickly in China, that is something NVIDIA should be most concerned about in my view. Yeah, I think that makes a lot of sense. And, you know, they're always thinking about how to partner more closely with researchers over there and, you know, what chips can they sell there even if they aren't the highest end chip. But that's a huge market for NVIDIA. So I think you're right. Right. What about the companies that they have invested in?

18:32We had a really cool graphic in the story that, again, we'll link it in the show notes. You should check out the story. We had a graphic of all the different startups that NVIDIA has invested in. As you studied that group of startups, what sorts of observations did you make? Yeah, so I think that NVIDIA has always invested in startups. So this is not something new, but it's definitely picked up pace as NVIDIA has grown in the past couple of years. And I think for me personally, one of the more interesting areas of investment for NVIDIA is these other industries that could be large, but today are not super large.

19:09And that to me signals that even though the AI market is huge, Jensen is very focused on the next big thing. So, you know, they got this huge boom from AI and LLMs, but that, you know, the company isn't satisfied. That's not where they're stopping. And so they're investing in other industries. So quantum is one example of that. We named a few quantum computing companies. And then I think another area that's interesting is just their investments in cloud providers. And so they've sort of picked up the scale of investment in cloud providers to really make sure that there are enough companies out there that can actually build the data centers that are needed for NVIDIA chips.

19:51Right. I want to pivot to talking about a story that you published this morning, which is about Sam Altman's ambitions to possibly build 250 gigawatts of compute capacity and power. Gosh, I don't even know where to start with this. It feels like we've heard something like, wasn't there a$7 trillion figure once that OpenAI had talked to? They want to spend that much on compute. I mean, is this like another outlandish figure that we're just throwing out there? Or is there any sort of backing to this? Yeah, I mean, there are definitely people within the industry that kind of roll their eyes at these numbers every time they're thrown out.

20:31OpenAI is expecting to end this year around 2 gigawatts of power. And so now they're projecting at least internally to employees, could we make that$250 in 2033? And that is like a major percentage of the entire output of the United States. I think we have a graphic in our newsletter that also compares that to other countries and even all of the nuclear power plants in the US. And so it is a crazy number. And I don't think we're taking it extremely literally. But in order for OpenAI to expand the amount of energy in the world, it's going to involve a lot of moving parts. So I think they are trying to project to the world.

21:13They're trying to project to the makers of gas turbines, you know, data center developers, utilities. I think they're trying to project this crazy expansion to get the whole industry on board because OpenAI won't be able to reach 250 on its own. And I think this is sort of a way to instill excitement in an industry that typically doesn't overbuild or do anything that's too risky. Right. And do we have any estimations around how much that might cost to build 250 gigawatts of power? I think we're safe to say that's in the trillion dollar range. Who knows what the cost will be by then, but it is just a number that's hard to fathom.

21:58Right. And I mean, last question for you, the numbers are so big. I mean, is it really just a fundraising tactic at this point to throw them around? I mean, it feels like, you know, why hold back your ambition? Why not? You know, that's what it seems like to me is, you know, you could say whatever number you want at this point, you just got to keep raising the money. Yeah, they're certainly having those conversations right now as well. And at least the people that I talk to are seeing these signs as like, you know, OpenAI must have a really good reason to need 250 gigawatts. So I think it's working.

22:31Right. Great. Well, Anissa, thank you so much for coming on the show. That is Anissa Gardizi who covers all things cloud and compute related here at The Information. Okay, the AI boom has led to surging interest in robotics, and humanoid robots have become a big focus for the tech sector. We have previously written at the information about just how hard it is to get hands on robots that mimic human dexterity, and a new essay from the co-founder of the company behind the Roomba and iRobot argues that today's humanoid robots won't be able to learn dexterity as well as we think. Joining me now is Rodney Brooks.

23:08He has a new company called Robust AI. It is his first time on the show. Welcome to TI TV, Rodney. It's great to have you. Thanks for having me here. What a wonderful essay that you wrote. Sorry, we'll link it in the show notes. I suggest everyone give it a read. Look, I just want to ask you, why don't you think robots will be able to learn dexterity in the way that so many companies today advertise they will? Yeah, and that's the point. I'm not saying they can never learn dexterity. I'm just saying people are not collecting the right data at the moment, and they're not trying to learn the right things.

23:44So what most of the companies are doing is using first-person video, looking at someone's hands, doing some task, and from that, trying to learn how to be dexterous. But our hands are full of touch sensors. Touch is completely part of everything we do. You know, you can reach inside your gym bag or something and pull the thing out you want. You don't have to look. And you use that sense of touch all the time. And since they're not learning from touch at the moment, mostly, and explicitly some of them have said so, I don't think they're going to learn to be dexterous like we are. And the whole argument for humanoid robots is you build one sort of robot, a humanoid, and it can do everything.

24:27So you don't have to build, as the CEO figure says, millions of different sorts of robots. So it's got to be pretty damn good at dexterity if it's going to do everything that you could ever want a robot to do. Right. And so you're not saying that we're not going to get to the point where these robots are built. You're saying that with the current approach, we've got to use touch much more than we are right now. Are there companies using touch at all? There's a little bit of it, and it's more in academia, but I think this is a mistake. And most of these robotic humanoid companies are being started by AI people who haven't worked in robotics.

25:02and they make a fundamental error. They think that what a robot does is produce trajectories where its arm moves in a particular way. No, what a person does when they interact with the world is they apply forces and they sense what happens when they apply force. It's about the contact. The contact is the important part. And that's what robotics has learned over the last 50 years. And that's sort of being ignored. You know, we see a robot move like this, we'll just make it move like that. And then everything is the same, but it's not the case. Robotics is about energy management. It's about putting energy into a system and, importantly, taking it out to preserve safety to be around humans.

25:41And that's another problem humanoids have. So your company, Robots, what kind of robots do you work on? We build carts. They look like carts. They actually have a top-of-the-line NVIDIA in the top doing vision and looking at the world. And we've got embedded processors. And they're for warehouses. and they go around in warehouses and help the human pickers. We don't try and do the picking operation. We leave that to humans, but we make their work much easier, reduce the number of steps they need to take, and it's a real productivity gain. It's a return on investment. And that's the other thing a lot of AI people and robotics people don't really quite know.

26:21I've got a great technology. Everyone's going to want to buy it. No. People buy things that are good for their business, return on investment. You've got to get return on investment. And you're not going to break into a market unless you are showing your buyers, at least your financial buyers. You've got different buyers at different levels in companies. You're not going to get them to buy unless they see a return on investment. And so that seems to be a significant challenge for human robots at all, is that they're so expensive to make and buy. I mean, we're still dealing with ROI with with, you know, simpler robots, I guess.

26:56Yeah, exactly. Exactly. And there's a whole bunch of things around it too. You know, when you, we at Robust AI, we sell into warehousing and manufacturing. So we deal with big companies, small companies, all kinds of companies. But in every case, we have three buyers. We have the company who wants return on investment. We have the manager of the local facility who, you know, he has to, he or she has to get stuff done every day and doesn't want disruptions, wants them to be entirely reliable. And then you have the human workers who are still going to be around for a long, long time. And they have to accept the robot as not being scary and being something that they get to control and they have use of rather than the big bad bosses providing this scary technology for them.