AI's Impact on Advertising, Enterprise, and Software | July 29, 2025

0:12Welcome everyone to the Informations TI TV. My name is Akash Pasricha. It is Tuesday, July 29th, and we have got a busy, busy show planned for you today. We have got Joe Marchese on the stream today. He's going to talk to us about AI and advertising. AWS is joining us to talk about the role of the chief AI officer. We have also got news about a new investment out of Snowflake Ventures. And finally, we are going to have our Microsoft reporter on to talk about how AI is lowering switching costs for software providers. That is a very interesting discussion I am looking forward to. I want to start with two great stories out of the information over the past 24 hours.

0:52Our Asia Bureau published a piece this morning about how AI coding tool cursor is becoming very popular in China. And one of the most interesting parts about this story is that engineers in China are trying to access another popular coding tool, Anthropix Claude Code. But the restrictions that Anthropix is putting on China customers accessing that program is making them turn to Cursor. And that's what's leading to an increasing popularity for that platform. It is a very interesting story. Another big story that we published was a scoop that Fireworks AI is in talks to raise money at a$4 billion valuation.

1:28It is yet another cloud company that is renting out NVIDIA chips. Lightspeed and Index Ventures are the names that are being floated as the leads. And we also got our hands on the company's revenue projections. So be sure to check out that story. Now, one more big headline before we get started. Spotify reported quarterly earnings today. Revenue was up, but shares were down. The most interesting part of these earnings is the fact that Spotify's ads business shrunk slightly this quarter. And Daniel Eck, the CEO, said on the call that he wasn't happy with the team's execution. It's kind of hard not to connect that with the fact that Spotify's ads chief left the company yesterday.

2:05we learned they haven't found a replacement yet. And so Spotify's ads team clearly has their work cut out for them. So we will see how that story shakes out. Okay, on to our first guest. We just talked about Spotify's ads business. Advertising in general hasn't gotten enough attention as big tech earnings get underway. And so we're going to double click on that with our next guest. I want to bring on our editor in chief, Jessica Lesson, who has a very special conversation planned with digital and traditional ad veteran Joe Marchese. They're going to talk about all the little noticed ways AI is disrupting the ad business.

2:45I am so excited for that conversation. Jessica, over to you. Thank you, Akash. And Joe, thank you for joining us. You know, your day job now is as the general and build partner at Human Ventures. But you know the advertising business. You've been head of advertising for the Fox Networks group. So thanks so much for joining us at kind of this wild time. You know, I wanted to have this conversation because we're in the middle of earnings. We saw Google last week. We have Meta tomorrow. It seems like we know how disruptive AI could be. Everyone is talking about it. But these ad businesses are holding up.

3:29So I want to talk to you about where you think we are, but maybe looking into your crystal ball a little bit, what we should be thinking about when we think about AI disrupting advertising in the coming months and years. Yeah, and thanks for having me. I think the first thing to level set is disconnecting AI disrupting advertising from Google, right? AI will disrupt search in some ways, but not search querying. Like more people are querying Google than ever. The question is, is traffic going out to the web as we know it? And so the web was built on traffic being sent to publishers and websites or travel sites or restaurant sites or recipe sites.

4:11And all of those or most of them were built on an advertising backbone. So it's hard to overstate. And people think of advertising as a small category. I'm almost surprised it's not covered as much. The largest companies in the world, like you mentioned, Meta, Google, Amazon, massive ad business, Uber's now in the ad business. It is Walmart and its retail media network, like drive its stock price. It's straight to the bottom line. And so what AI disrupts is the flow of search. And after Google, YouTube is the next largest search engine. And after that, you have Amazon and Walmart and the retail media networks.

4:46And if agents are acting on the behalf of humans rather than searching through and having a chance to advertise to people, you end up with a disruption of multiple, multiple billions of dollars. Yeah. So obviously it's early. We're not seeing it yet. Or are we? I mean, what do you see when you look at the results Google put out or what you're starting to see in the products? Well, I think you are seeing the impact already in terms of traffic from Google to publishers. And the irony there is that publishers are the things that large language models need to train on. And that is why you have, if you haven't seen, Matthew Prince gave a talk out at CanLion.

5:25So why is a Cloudflare CEO at CanLion without an advertising business? And it's because the amount of times that LLMs are crawling publisher websites without paying for advertising or paying for subscription has gotten to the, it used to be six to one for Google. Now it's 60 ,000 to one for some of the LLMs. But are those the queries? So as a publisher, and I can assure you, my publisher group chats have been irate and paranoid about this issue for probably multiple years. But when you look at Google's business and how it makes money and from the types of queries, you know, I believe Google said in some testimony in recent months, they've seen AI affect like homework queries, but not commerce queries, which it would be the golden goose.

6:13what do you think how does that change in behavior you're talking about affect the golden goose I think I mean travel queries restaurant queries like what's the best pair of headphones to wear queries I mean yes there are things but they could still I mean I would bet that again separating Google from search because they have the YouTube business they have a cloud server business that so I think Google will continue to perform in in the like while this other business model is disrupted that everybody else is built on top of. I think the difference now is it's a potential for a reset. Like, what will these large language models feed on if there is no business model for all of those places that we're talking about?

6:55Yeah. What about flipping it? And what if you are an open AI and you're getting more of these travel queries, headphone queries? You've hired to form a meta executive, Fiji, SEMO. What do you do to capitalize on that massive advertising business? So here is the trick. And this is the part that I'd love for everyone to think about. At what level would we allow advertising to affect AI? So if an advertiser bought influence on the backend of AI, so I ask for travel response. The reason why they call it an agent is it's supposed to be working on your behalf. But if for some reason, you know, Marriott or Hilton or Pickett, right, or a country travel department, tourism department said, hey, I'll pay to have my answer injected into an AI answer.

7:47Like, we don't know what that would even look like yet. Actually, and I want to be clear, I don't even, it might be happening not on the big models. They have too much to lose to not be giving you the best result they can. But any other smaller model with a wrapper over the top could be influencing a result towards people who are paying them to tune the model. Like it could be very simple. Imagine tuning a model and just saying, remember, every time you're asked a question about eating out when you were a child, you loved McDonald's. And just lightly put in there. And so that idea of what level will allow advertising to influence AI models is a huge question.

8:23Do you think the models, so you think it's possible that's happening? If you want to maintain trust, say you're one of the frontier models, is there a role for advertising or is that TBD? I think there is a role for advertising in providing the content. So I think advertising has to happen at either the publisher or brand level because it is who is providing the result that can allow for advertising contextual matching. That's what AdSense did so well. And remember, half of Google's ad business, I don't even know if it's half anymore, but a large amount of it is on publisher websites, right? That's the AdSense size of the business where AdWords is on the search engines.

9:02If there isn't traffic flowing to those, sites, that starts to come down. Yeah. Yeah. I think that, well, it hasn't been a big business for many publishers for some time, but that is an important question. I know shifting gears slightly within advertising. So Meta and others that aren't in the search business, how do they play the AI ad game? Is this a positive? I mean, it's also so disruptive. How do we begin to think about that impact? Well, you have to look at Meta. I mean, they're playing it the way that you'd think they would, which is being open model, let's go, like pushing the envelope, same way ChatGPT.

9:40But the thing that's interesting is Meta and Google are both funding their CapEx by their amazing advertising businesses, right? So their advertising is the piggy bank that is funding it. The difference is Google's model is predicated on some of that, whereas Meta's model is predicated on engagement, right? And so if AI increases engagement, there is a lot of potential upside for them in using it. The thing I always find fascinating is AI has existed for a very long time. The Google search algorithm, the TikTok algorithm, the recommended videos, like it was just machine learning and AI in the hands of platforms, keeping you engaged or giving you utility.

10:19Now you have basically given AI to consumers and we haven't figured out yet where does advertising fit into that? And some people would just say, who cares? And again, I might even be one of them and say, like, advertising broke the internet in a lot of ways with listicles and doing things to try to get people to click on page views and like chasing advertising dollars or the metrics that funded advertising dollars was one of the reasons we ended up here. I could make the same argument that advertising broke cable. Like the entire cable bundle would have been fine with the better user experience over time because we're getting back to it.

10:53Different conversation. But free is powerful, man. Free is powerful, right? It's just really hard to imagine the AI ecosystem without advertising. Right. And well, or even simpler, just look at if we remove tens of billions of dollars from the ecosystem, as little as it might be that's still flowing to publishers and content creators, if we remove it, then that is just less money to be shared amongst people who write content, who create the stuff that then feeds the model. So that's the question I have is like, this isn't a next year, two years thing. This is the next six months if at the rate adoption is going.

11:31Yeah. Well, my last question for you, you're a startup guy now. You see probably the most interesting startups across media and tech. Are there opportunities for founders to catch this wave, to be early adopters and thinking about what the ad model looks like or much like the model development? Is this really for the big guns with little room for the small guys? No, I absolutely think there is. I mean, if you think back to search, Overture, like click, pay-per-click marketing wasn't just invented by Google in up north. It was actually out of LA where someone said, hey, we have an idea here. I think that looking at AI and saying, okay, if consumers are going to interact with an agent that's on their behalf, what role would advertising play?

12:16The things I think to think about I'm seeing companies that are structuring MCPs, which is basically designing websites for computers to read instead of humans to read. And you could build into that. OK, well, you'll read the advertising as well. You know, there are PR firms. SEO seems to be one of the most important things. The problem is no one knows how to value it. Like getting mentioned in an article means you're more likely to come up in a result in ChatGPT, which is ironically running on top of the Bing crawler, which very few people have paid attention to. And so we've had those discussions and I see interesting models all the time.

12:49So if that layer becomes not commoditized, but ubiquitous, that AI can be plugged into everything, then the question of what an ad experience looks like is actually fair game. Yeah. Well, I think that's a great point. We need to now turn our attention to what layer. Is it the agent layer? Is it the site layer? And I think this Wild West phase will certainly yield a lot of opportunity and eventually disruption. So thank you, Joe. You've given us a lot to think about. Good to be here. And back to you, Akash. All right. Well, our next segment is with our presenting sponsor, Amazon Web Services. So far on the show, we have talked a lot to founders and investors who are building AI for businesses.

13:35But the thing that doesn't get talked about enough is the fact that businesses can't just buy and switch to AI overnight. In fact, for a lot of these businesses, I'm talking about automakers and retailers and financial institutions, they don't even know where to start with AI. And so I want to bring on someone who helps people with that issue every day. Sean Nandy is a director at AWS. He is the former CIO of Dow Jones News Corp. And I am really excited to have him on. Sean, welcome to TITV. Thanks for being here. Gosh, thanks for having me today. Super excited. So look, I want to get into the meat of the matter here, but very quickly, tell us what your role is at AWS so we can set the stage.

14:14Absolutely. So I've been here almost six years and I look after what we call our field tech teams. These are all of our solution architects and customer solutions and success folks who help our customers build with AWS. So we go out there, we explain our services, we understand their business challenges, and we think about how they can use both cloud and AI to get to real outcomes. So the thing that I want to talk to you most about today is talent, because the biggest story in AI right now today are the massive sums that are being spent on hiring talent at tech companies and fast-growing startups.

14:48But the other half of the story is that established companies, you know, companies that have been around for decades, they're trying to build out their AI strategies right now. And in my mind, it's a very significant transformation. So, I mean, let's just start with a basic question. How is that shift going for you right now and the people you're talking to? You know, it's a constant conversation and I have that conversation at every level. I'll have it with boards, I'll have it with CEOs, I'll have it with CIOs and CFOs, even line of business leaders. And the conversations are super nuanced. It's both on what skills do I need in my knowledge worker population to leverage all these tools and think differently with AI.

15:28But the first one, who's going to lead this charge for me? Who's going to really lead and drive AI adoption and think about the guardrails and what partners I use? And how do I staff that role? So we're increasingly seeing a role of chief AI officer, sometimes it's chief data and AI officer get sort of built out in these companies. In fact, I'll give you a stat. We just published a generative AI adopters index a few months ago, and we interviewed, I think it was almost 4 ,000 decision makers at different companies, not AWS customers. This was super broad, everyone, every type of customer. And what we found is as of today, 60 % of enterprises have named some type of chief AI officer.

16:08What's even more astounding, another 26 % are applying to do so in 2026. So 86 % of companies are going to have this sort of nebulous, ambiguous chief AI officer, chief data and AI officer. So what is this chief AI officer? What does that person end up being? That's almost the first conversation I get to have at these board level conversations. Like, who should we hire? or maybe I get to talk to one who's newly appointed who's like, what should I focus on? And I'll give you a little bit of a track here. First of all, every enterprise is at a different stage of adoption with cloud and with AI. And those who have been very forward looking with technology the last decade or so are often in a better position, right?

16:51They've been maybe experimenting with AI. They've bought some early tool sets. They've done some implementations. Others, they've maybe avoided the journey to cloud. They've been nervous about it. They've been waiting for the right inflection point and they're worried about being behind. So the folks in this role are quite variable. Some of them might have been a data scientist or an ML engineering leader who was already in the company. Sometimes it's a former CIO. I've even seen line of business leaders asked to take on this mantle. And so depending on the persona, they come in with different expectations and perspectives.

17:23Sorry, a little technical glitch. They come in with different expectations and perspectives. And when I talk about the different expectations and perspectives, sometimes they'll come in and say, We want to govern security and compliance. Maybe it's a former CISO. We want to make sure that we're safe. And even that safety perspective, it's not just about data leakage. It's about legal exposures. On the other side, something was like, we're in with procurement. What vendor are we buying? What vendor are we going to buy software from? How are we going to govern it? Those are places to start, but not where I would start.

17:56So where'd you start? Yeah. I think the real thing you want to make AI adoption matter in enterprise is to democratize it, to get the line of businesses in and thinking about not just how they can save money, but how they can innovate faster. That means you have to have representation from those groups. And when I think about representation, that means it's not just about having a single chief AI officer, it's that team they can harness. So we encourage them to really build a center of excellence and to have that center of excellence, which doesn't have to be a bunch of people, by the way, it can be a big virtual, have that center of excellence offset the skills they're great at.

18:31So if you have a CISO running, becoming this chief AI officer, go get an engineering or product leader to join that extended team. If you have someone who's a former CIO, get some security folks in and think about how you want to show what success looks like. And, you know, just to, you know, simplify things here, but when you say center of excellence, you're basically talking about, you know, a team of executives rather than just one person. Is that the idea? It is. And by the way, executives are important in this, but like AI is super powerful. You really want some doers who can actually use the technology and start to harness the power of it, but also bring people on the journey.

19:10Because of course, the knowledge workers are reacting to this and saying, what's this mean for me? What's this mean for my job? And so I think this chief AI officer also successful if they're a change agent for the company. When I've had these CXO discussions, I was just in Detroit with one of the CEOs of one of the big three automakers. And I was with their staff, and they were asking about how AI might change their business. It was a super exciting conversation. And our conversation all came back to people. When we get people efficiency gains and innovation gains and changes, how does that value come back to the enterprise?

19:45What keeps people from just going home an hour early? What does it make it actually bring impact to the business? And how do you get them in? And we were looking at their chief AI officer, who happened to be a highly technical cheap AI officer who could really talk about what model was the best model and helping think about how product engineering could be improved, but they weren't thinking about the change management and people aspect. So we're like, you need a really strong line of business leader on this team who can really think about if their people are going to work differently. So, you know, I think it's kind of interesting that, you know, the point I'm taking from you is that it doesn't necessarily need to be the one person.

20:20You need to have the team underneath. They need to have oversight of different parts of the business. I mean, I guess the thing that I'm wondering from you is, does everybody need a chief AI officer then? I mean, you talked about that stat that everyone's naming this person. It strikes me that maybe you actually don't need to name it. It's more of sort of an organizational shift. Look, let's think about where we want to get to versus where we are today, right? Like five, six years ago, you'd see companies start naming chief cloud officers. thinking about cloud being such an important part of the transformation.

20:55And now cloud is core part of a CIO or CTO or CFO's job. It is not a separate thing. I think having a cheap A-opster is an acknowledgement that this is a major area of change. Someone has to lead the charge. But do I think it's a title with permanence? They're not going to be AI companies in a few years, right? Every company is just going to have AI embedded in it. And that's a fundamental change. So whether you name a dedicated role or you simply acknowledge it's impacting all the existing roles, it's that journey there. I do think for enterprises, having a cheap AI app store can be a great way to sort of crystallize to everyone.

21:29We're jumping on this. We're embracing it. The skill set, it's not a given. And by the way, you need technical chops to work in this space, but you don't have to be a data scientist or an ML engineer to set company strategy and to evaluate partners and to think about what has the biggest business outcome. So having a relative amount of business acumen, and then I'll give you the most important characteristic in these types of roles. And at Amazon, this is a core leadership principle for us. We have these leadership principles in Amazon that help us think about how to operate as employees and builders in our own organization.

22:02One of them is called learn and be curious. And if you're looking for a chief AI officer, you need someone who wants to work with and use the technology in their everyday lives and at work day in and day out. They have to be passionate about learning because things are changing so quickly. And I want to just jump in on that point. You know, things are changing quickly. This role that you're talking about, how has the role changed in the past year? And really, what do you think this role will be three years from now? Yeah, look, I think in the last year, the change I've seen, again, enterprises, mid-market companies, they're all at different paces in their rate of adoption and where they started from.

22:41But I'd probably say the role started with people looking for someone with those deep skills. They're like, let's go to one of the big AI companies like us and like our peers and hire someone who has these magical skills who can explain it to us. And I think companies are realizing it's more about knowing your business and being able to learn about it. That's part A. Part B was probably companies like, how do we build our own models? Do we need to? and now they're realizing, no, in most cases, we need to pick the right partner that can take us on the journey. So how do we govern and manage those partners and where do we apply that tech?

23:17And last, getting the whole company leadership team on board with the journey. So thinking about like, where do we want to apply it first? Which lines of business will have the most impact? Is it really an IT spend issue? Probably not. This is really about business change and business spend. And so getting a lot of business leadership. And what about three years from now? Where do you think the role is going? Yeah, I think the role in three years will probably be more of a visionary and governance role. By the way, visionariness and governance we think about is pretty far apart. But I think in three years, AI will be so embedded across lines of business.

23:52It'll be so embedded across vendor and partner tool sets. It'll be in everything that this role will really help security and compliance think about how do we move fast to adopt. It will help knowledge workers and leaders be inspired by what's possible with the technology and continue to change. I think that piece will persist. I think the actual guts of running it, that'll be super democratized across the business. You will not have separate teams just for AI anymore. You talked about that example of the automaker that you were visiting in Detroit. Right. You know, can you think of any other sort of, you know, customers that you've worked with or really interesting case studies where the role of the chief AI officer has really taken on sort of an alternative shape and it's worked really well in ways you may not have expected?

24:40Yeah, look, I spend a lot of time with every type of customer and it's very different in our deep tech ISV partners versus, say, financial services or health care travel. But I'll give you a couple. I think in financial services, working with banks, we've seen much faster adoption of AI than we expected. So, you know, a pundit on the side might say, of course, you know, financial services companies have large budgets and they have large tech teams, but they have often been slower to adopting technology because of regulation and compliance. And FinServe companies, they knew their data. They were able to decide really quickly where they wanted to apply their data in.

25:16And so we've seen chief AI officers in those companies really look at those success stories and replicate them across other lines of business and help broaden that thinking. I think in travel and hospitality, we're seeing a lot of pressure. Post the pandemic, you had a lot of turnover in these industries. A lot of frontline talent was lost. And so there's a lot of hunger to think about how do we use AI to help customer experience really get better, combined with, of course, all the pressure from the macro and economic environment on saving money. And so cheap AI officers, I think, are really navigating helping companies decide, am I trying to save money?