AWS’ New Enterprise AI Push, OpenAI Deal with Broadcom, Big Tech AI Ad Planning | Oct 13, 2025

0:12Welcome everyone to the information's TI TV. My name is Akash Pasricha. It is Monday, October 13th. Hope you had a great weekend. We have got a big show planned for you today. OpenAI and Broadcom are announcing a major chip deal over the next four years. Our cloud reporter will be up very shortly to put up the news in context. The chief marketing officer of Amazon Web Services is coming on in just a few minutes, so do not go anywhere. I'm excited for that conversation. We are also talking about the tokenization of real-world assets, which frankly is something I'm really trying to wrap my head around.

0:45So I'm excited for that chat. After that, we're going to be talking about the AI talent wars with someone who has built a big business in the land of executive search. And finally, we have got a great story about how Salesforce, Google, and Microsoft are planning their AI-powered ad tech push. There is a lot to get to, so let's get right into our first guest. OpenAI and Broadcom are teaming up. They announced a plan to roll out 10 gigawatts of custom AI chips over the next four years. Here to put the news into context is Anissa Gardizi, our cloud and compute reporter here at The Information. Anissa, it's a busy morning.

1:21How are you doing? I'm doing well. Happy to be here, Akash. So let's talk about another chip deal that OpenAI has done with another chip company. Give us the details of what we know, and then we'll talk about the bigger picture here. This morning, OpenAI and Broadcom announced a major partnership to deploy 10 gigawatts of custom AI chips for OpenAI. And it was broadly known within the industry that OpenAI was working with Broadcom on its own chip effort to help reduce its reliance on NVIDIA. But I think the company is publicly talking about it today and putting a figure on it, 10 gigawatts over four years, really shows how serious the two companies are.

2:05And OpenAI does need to raise a lot of money for this effort. So I think we should expect maybe the next couple of Mondays we might be on TITV talking about them. And it's kind of funny because over the last couple of weeks, as we've seen OpenAI do a deal with NVIDIA or NVIDIA announcing its letter of intent to invest, more money into OpenAI. Then we saw the OpenAI AMD deal. I mean, the question that we kept asking folks is, well, what does this mean for OpenAI's own chip efforts? Because we knew that they were going to be working on their own chip. And I guess this is sort of the closest look we've had at saying, yeah, we're doing it.

2:40And we're doing it in a big way with Broadcom as well. Right. I mean, 10 gigawatts really isn't a joke. And that's the exact amount of the strategic partnership between OpenAI and NVIDIA. Of course, OpenAI expects to get NVIDIA chips through other deals. But, you know, given that this now internal chip effort is the same size as the strategic partnership, I think it really puts into the context how much OpenAI does not want to be reliant on one company or even two companies. They sort of want many companies helping them reach their compute goals. And Jensen can't be happy about that. My guess is not really, especially given the way that this was announced today.

3:22I think the main takeaway was that OpenAI and Broadcom are not just working on a chip. They're working on the entire data center system, including networking and other things you need to run a system. And that is sort of NVIDIA's pitch. They always say that we do more than just building a chip. And so today, that is kind of what Broadcom announced as well. And this is one of the things that we've talked about on the show is that Broadcom has sort of some of those products that help build out sort of that full data center infrastructure. It's not just the chip itself. Exactly. And I think that's probably why OpenAI was drawn to a partner like Broadcom.

3:59Of course, we'll have to see how this all shakes out. We haven't really seen any examples of these Broadcom chips in action yet. So, you know, with all these announcements, we're going to have to see what happens. But the intent is very clear. Right. As you're looking for this story to develop further, what are you watching for? Well, one thing that we noticed this morning was Broadcom's president was on TV this morning talking with Greg Brockman. And he was asked the question, oh, well, is this your$10 billion customer that CEO Hawk Tan talked about on the last earnings call? And he kind of made it seem like OpenAI was not that$10 billion customer.

4:37So definitely one of the big questions today is, is that true? And then who is the$10 billion customer? Because Broadcom might have another relationship that they aren't publicly talking about that is still going to be important. Right. Well, Anissa, thank you so much for coming on short notice. That is the beauty of breaking news. We look forward to having you back on as we see more of the developments in the story playing out. That is Anissa Gardizi, our cloud and compute reporter here on TITV. Okay, Amazon Web Services made big news last week with a new suite of enterprise-focused AI tools that could challenge the traction of fast-growing enterprise search companies like Glean.

5:14Some of AWS's tools are new, but some of them are also kind of a repackaging of older AI tools it has already had in the market. Of course, AWS also has a lot more under the hood with its cloud business and its own chip efforts, but never before has there been more competition on both those fronts. And so I want to bring on Julia White, the Chief Marketing Officer of AWS. It is her first time on the show. Welcome, Julia. It's great to have you. It's so great to be here, Akash. Thank you. So I'm excited to talk about some of the announcement you guys made last week. One of the fundamental questions I do have is you guys unveiled QuickSuite.

5:46How is QuickSuite different from Q, which is another product that we talked a lot about here at The Information? Well, QuickSuite really builds on the learning we've gotten from Q for business, as well as QuickSight, as you also alluded to. And then it really germinated from watching our internal employees and the tools they were using and the needs that they had. And we brought all that together. And as you know better than anyone, right, the space is moving fast. And so our ability to really watch, learn, pivot, and adapt and bring them into a cohesive experience, as we call Amazon QuickSuite, is really the evolution that we're bringing to market this week.

6:21So does Q fit inside QuickSuite? Or how does that work? It's really the next generation. So our QBusiness customers are being evolved into the QuickSuite application that takes over those capabilities. Of course, if you're on QBusiness, as we do with any enterprise, we continue to support that. But really, it is the evolution of that, and it will replace it in the fullness of time. Well, one of our reporters wrote about it last week, and one of the things he mentioned is it felt a little bit more like a rebranding than an introduction of new products. Do you not see it that way? Well, we bring a lot together that is brand new.

6:56For example, we have industry-leading quick research, which is one of the capabilities in it, where you can go deep, deep, deep and understand what's happening across the whole industry sector, put a whole strategy together for you as an example. But then the entire user experience and the application experience itself is new and kind of, again, germinated out of watching our internal customers as well as our external customers and bringing those pieces together. One of the neat parts of QuickSuite is sort of the enterprise search tool and the ability to use this product across all of your other software that you have installed.

7:32It's something that we've talked about in the show in the context of fast growing startups like Glean. One of the challenges that we've written about here at The Information is this idea that some software companies may block or may throttle access to the data that's hosted inside their own programs from AI companies who are looking to access them. This is the whole issue around corporate data wars. I mean, how much of a challenge is that for you right now? Well, we've had in market our index, it's now part of the quick suite again, quick index, that is enabling customers, ourselves and others to tap into things like Salesforce and Slack and third party.

8:09And obviously all of the data sitting in AWS, which is where most of the world's data actually sits. So we have access to all of that. But then the new aspect of it as well is also your own personal information, like Microsoft 365, your docs, your PowerPoint, your email. And we don't see in terms of we're using public APIs, we're working in partnership. with these organizations. And so to your point of, we don't see that on the horizon, but understand the concerns. Understand the concerns. And so, but you're saying that you don't foresee it being an issue, even though it could be one way it goes.

8:43I think we're doing it in partnership with so many of these companies. Again, and so many of them are customers and partners with AWS as well. And so very much doing it in partnership with them. Right. I want to talk about the ROI that some businesses are trying to get out of AI. It's something that has been difficult for some businesses, given that these projects can be expensive. And also, you might not know the right ways in which to use the technology. And agents are specifically one area of AI where people have kind of been scratching their heads saying, hey, I'm trying it. And it's slow.

9:16I mean, you know, I'm not seeing the impact just yet. Are your customers actually seeing ROI? Or are these all just sort of pilot projects that they're just playing with the technology right now? But we've definitely seen the kind of what people call it, the pilot purgatory, right? And I think that's why we made the learnings we've had, again, that we've been able to watch, learn, pivot with QuickSuite. The getting up and started is like zero friction, right? So as a user, you already have your data set up. They sent your organizations to put that in place. But you can also just immediately bring in, again, your documents, your access to BI dashboards, even information sitting in like Snowflake.

9:51And so the user can get moving very, very quickly, which can get you out of just I'm stuck in pilot and I'm actually can use it in a real world way. Right. These tools get access to all the business data. You're not just trying to work around it, which is usually the problem of getting ROI. So, again, just within my own organization, we now run every marketing review using QuickSuite because a team can pull everything together instantly. We know what kind of insights we're looking for, and then you can put it into a review for all of us to share. So then that's you're now measuring from an ROI perspective.

10:20Okay, if my team now takes half the time to get their business review pulled together, what's the value of that? And their ability to go instead work on another marketing campaign using those insights versus that two hours to actually just find the insights. Right. But as you're talking to customers, what is the biggest challenge that they have? Or what is the biggest factor that is preventing them from spending more on AI right now? Mostly it is about, it's interesting, it's such an old world problem, but about data silos, right? And that's what slows down the benefit of so many of these tools is getting access to the right data, bringing it together, making it useful.

10:54And so, again, one of the things we spent so much time on was quick index, making sure that the data silos could be harmonized. And again, take a different approach, right? A lot of the other people in this space are doing it application-specific or data-silo-specific kind of agentic experiences where we looked very broadly. taking from our history of being AWS and a broad set of infrastructure and data systems that we could bring that forward into the application layer as well because that's what slows people down and that's what makes it not as useful and where the ROI struggles when you can't get access to the right data.

11:27Right. AWS has its own tools to help people develop their own agents. We also saw last week OpenAI unveil a similar tool in that category. What did you make of their release last week? I think, you know, the agentic space is moving incredibly fast. I think we're going to see a lot of experimenting and then continuing to evolve. And as you've seen us bring out, you know, Kiro, our agentic IDE for developers and seeing really, really wonderful success in terms of adoption with developers and what they're doing with it. Now with QuickSuite for all of the employees and kind of users at organization.

12:01And then even just today, our new agentic platform called Bedrock Agent Core is in GA. So all the different tools, whether you're a developer, whether you're building out the whole infrastructure for Argentic Systems, or you're an end user, covering all those different bases. Right. And one of the other things that we've written about here at The Information is sort of the margin pressure that comes up for cloud providers, given how expensive some of the underlying technology can be for renting out these AI chips. Are you guys seeing any of that margin pressure in your own business? You know, our business continues to be quite healthy.

12:36And as you see in our earnings, I kind of the way we look at it and report, it continues to be quite steady. And what about the Tranium chip that you guys have? It's an interest. I mean, we were just talking earlier on the segment before you about OpenAI's own ambitions to develop their own chip. It's something that we were waiting for more details on. Of course, Amazon has had this effort going for a long time. Can you talk about the extent to which AWS customers are actually asking to say, hey, we want to use that chip or we want to use the compute associated with that chip instead of something like NVIDIA, which they might have been used to doing for the last two or three years.

13:17Are they actually making the ask to transfer right now? As you said, I think our investment, our long-term investment in our own chips, with Tranium being kind of the most current example is really a wonderful advantage for us, right? From a price performance perspective. And so if you just look at our customers using Bedrock, right, our AI platform, over half of that is powered by Tranium. So over the half of customers of many, many that are using Bedrock are already using that as well. And then our big, big investment called Project Rainier, our super AI kind of build out with, in partnership with Anthropic, also powered largely by Tranium as well.

13:51And so I think proof's in the pudding in terms of seeing Tranium payoff and the price performance we can provide. Of course, by the way, we're wonderful partners with NVIDIA. We host more of their chips than anyone, but it's an opportunity for customers to have variations. Right. And I mean, I just wondered if internally the company had a goal in saying, I think I just missed that detail. You said half of Bedrock customers are using? Yeah, over half of Bedrock is powered by Tranium, right? So by definition, and then half the customers using FedRock are powered by Tranium. Right, great. Well, Julia, I want to thank you for coming on the show.

14:28It was an exciting week of news last week and it very much is something that we've been interested to see how it plays out. Thank you for coming on the show. We appreciate it. And as you have more news, please come back on the show and tell us more about it. That is Julia O 'White, the Vice President and Chief Marketing Officer at AWS here on TITV. We have talked a lot on this show about ways in which AI is changing the advertising landscape It also means big changes are coming for the ad tech sector. Today, the information published a story not just about how existing ad tech giants like Google and Amazon are using AI, but also about how other big tech companies like Salesforce and Microsoft are getting into the game.

15:08I want to bring on Catherine Perloff, our advertising reporter, to tell us more about the story that she wrote. And also joining us in the conversation is Robert Webster, CEO of Tao Marketing Solutions. Catherine and Robert, it's great to have you both on the show. Welcome to TITV. Yeah, thank you. Good to be here. So, Catherine, let's talk about this story that you wrote. You know, one of the things that we've talked about is sort of the marketplace for ads all together. And this was kind of an interesting story because this is a story one level up. It's about how companies are sort of planning and evaluating these advertising campaigns that they do.

15:43And so tell us a little bit about some of the challenges that exist with respect to people planning these campaigns. I mean, you've got so many different platforms. What are some of the challenges? Yeah, you know, so the system for buying ads is very complicated. There's a ton of different technologies involved, especially if you want to buy ads on the web or on TV. There's a whole different system of different ad technologies you have to use. The other challenge is, you know, it's typically seen as a little bit easier to buy ads on like Meta or YouTube. but it's very hard to compare the effectiveness of campaigns across those channels because, you know, Google might use different metrics or count things differently than meta.

16:27So it's very hard to sort of know where exactly you should put your dollars and how well the campaign worked, which then is sort of just like a loop because, you know, if one campaign worked, then you should spend more money in that channel. So it's very hard to kind of plan across the top. And AI, you know, agents can help sort of simplify some of that potentially by creating an easier workflow and being able to be smarter about making decisions about the data. But one of the interesting things that you reported is despite how many big tech companies are in the advertising game, there actually seems to be a movement here where even more tech companies, companies like Salesforce that I have never associated with advertising, they are also looking into getting into this space.

17:16How are they looking to do that? Tell us about what you found. Yeah. So Salesforce, to be clear, has been in the marketing space before, but it's been more sort of like, I mean, they've had some media analysis capabilities, but their kind of main bread and butter that most marketers know them for is email marketing. and then like storing your data, which is obviously very related to their core business of just like storing customer information. And I think that, you know, with their new kind of, they have this agent force technology that they've been pushing recently. And with that technology, it doesn't just analyze which campaigns are working and which aren't, and maybe do that in a more sophisticated way, but it also can take an action for you.

17:59So if your Google campaign isn't working, it can go ahead and like stop it. And I think this kind of technology, like it's kind of jargony, but there's kind of been two different sets of tech in the industry, MarTech and AdTech. MarTech being sort of this more, just sort of storing your information, maybe your customer information, you know, sending them emails, sending them texts. And I think with AI, that distinction from AdTech, which is more about executing on media, is sort of collapsing a little bit. And I think this is a story we're seeing across enterprise technology that AI is sort of bleeding these distinctions between different types of companies.

18:40And we're seeing that kind of in the advertising space as well, which means more different players can get involved. Robert, how are your clients thinking about using AI in the context of this story? Yeah, so I think as Catherine said, it's about how you bring data together to make more holistic decisions. And a lot of it is around how do you drive efficiencies? Yeah, how can you remove a lot of that busy work that has to happen, that takes a long time? And how do you actually use your talent to do more creative, better work? So the objective here really is definitely an efficiency saving on time, but also driving better results.

19:19Yeah, we mentioned people like Salesforce, people like Adobe getting into this space. when you've got customer data also being used to make advertising decisions, that's just going to drive better results. Right. In a very safe fashion. And I'm curious, one of the things that we've been talking a lot on the show is the ROI that comes with using AI and agents. And it's something that businesses are really trying to find, really, given that AI, in many cases, can be very expensive. and the cost savings, the revenue uplift, it can be a little bit vague in some cases. Are you seeing, you know, really tactical ROI that is coming from using these agents in the advertising space?

20:03Or is it still sort of, hey, we're just trying to figure out how to use it in the first place? No, no, the ROI is here now. You've got to think, the agents have already really been ready for about 12 months. And marketing workflows are quite specialized. You know, it's kind of bespoke sort of skill set with some awful lot of industry jargon and in the past required experience. So if agents have been ready and available for a year, it's probably taken most of this year to get them used. But you're now seeing early adopters getting big returns on efficiency, but also ROI on this. Talk about how.

20:38Where is that ROI coming from? Well, so if you try to make absolute decisions now, you can make better decisions faster. In a world, let's say you try to plan all the media you want to buy. I haven't met a planner yet who knows TV, Search, Meta, TikTok, all equally well. And Equal is able to use all of the data sets equally well across the top. They'd also make that global. Let's say you make a decision globally about how to run all your media in like 50 markets. No human can do that. But an AI with a great human alongside them can do that in 10 % of the time to a high degree of accuracy. Catherine, I want to end with you here.

21:20Google is a giant in this space. Second to Google, who is the company closest on their tail right now? And talk about that ROI point that Robert was talking about here, because I think this is, it's important. If Robert is saying, hey, the ROI is here, this is actually something, it's an enterprise use case of AI that we haven't really talked about as much. You know, it might be the low-hanging fruit, something that we should be paying more attention to. Yeah. You know, Google, I guess in ad tech, quad ad tech, you know, Google probably is the leader. Amazon has been catching up to them this year.

21:54It's been a priority. There's some other smaller firms that are well-known and well-used in the industry, like the Trade Desk. But, you know, in terms of this sort of new area for media planning with AI, I think that this is a capability that hasn't really been as much a focus for tech, maybe. It's maybe been like, you know, I don't know, Robert, you know, what you think here, but like, you know, it seems like it's been something more kind of like that agencies and ad buyers would just do themselves. So I think it's just adding a new layer where tech companies compete. And I don't think that there is a clear winner right now.

22:26That's why it's kind of exciting to see. Um, in terms of, um, return on investment, I mean, I think that like, yeah, the, the sort of like classic marketing adage that, uh, is, has been said throughout history or I don't know, maybe the past 20th century is like, you know, I know 50 % of my budget work, but I don't know which 50%. And I think that like, there was a promise of technology to solve that because, oh, now we know who's looking at the ads and we can trace that to the purchases. It's honestly made it a lot more complicated because there's way more channels. People aren't just watching one TV show anymore.

23:03Like the media ecosystem is really complicated. The data is really complicated. And I don't know if we're actually closer with all this technology to answering that question, which ad actually drove spend. And I think there's a lot of wasted spend going on right now because marketers don't really know where to put their money. So there's a lot of potential for gain if we can really answer this question. I mean, something that I think is a little bit like, how do you really measure if something is persuasive? Some of it might not really be possible. I think some of it's kind of outside of the realm of science and technology.

23:33But I think some of it we could really measure a lot better. And I think that's why this is interesting. Well, look, it's a fascinating space and it's an application of AI that I don't think we talk enough about on this show. And so I'm excited to have you both back on in the coming weeks and months as the story develops to talk more about this. Thank you, Catherine. And thank you, Robert, for joining us. That is Catherine from The Information and Robert from Tau. marketing solutions here on TITB. Okay, when the crypto sector isn't talking about stablecoins these days, real world tokenization is another idea that has everybody energized.

24:07We're going to get into what this concept actually means in a second, but more and more companies have started to build products in this category, not the least of which is Robinhood. Securitize is another company in the arena. The company is reportedly considering going public through a SPAC at a $1 billion valuation, according to Bloomberg. I want to bring on founder and CEO Carlos Domingo to talk about where he sees his business going. Carlos, welcome to TITV. It's great to have you. Yeah, thanks for inviting me. It's my pleasure to be here. So I'm excited to talk to you about this business of real world tokenization.