Behind the Numbers: Google’s Plans After the DeepSeek Announcement and How Much GenAI Chatbots Can Erode Its Search Dominance in 2025

Audio by Marcus Johnson, Jeremy Goldman, and Evelyn Mitchell-Wolf | Feb 18, 2025

TRANSCRIPT

Marcus:

This episode is made possible by Zeta Global. Do you know what it takes to transform marketing into a data-driven profit center? Are you able to align the C-suite around your AI vision and strategy? Zeta Global has a playbook to help you get started. Download Driving Growth in the AI Era today, link in the show notes. It's free.

Hey gang it's Tuesday, February 18th. Jeremy, Evelyn and listeners... Something. I hope you had a good long weekend is what I was going to say. President's Weekend, right? Did anyone? No comments? All right.

Evelyn Mitchell-Wolf:

To what?

Marcus:

I hope you had a good long weekend. I thought you'd-

Evelyn Mitchell-Wolf:

Oh.

Jeremy Goldman:

We did. Thank you. How was yours?

Evelyn Mitchell-Wolf:

Absolutely. Mm-hmm.

Marcus:

No, moment's over. To people listening, I hope you had a great one. Not these two. This is an EMARKETER video podcast called Behind the Numbers made possible by Zeta Global. I'm Marcus, and today we'll be discussing Google with the two folks who refused to talk to me, so we'll see how this episode goes. One of them is called Jeremy Goldman, and he is our senior director of briefings living in New York. Hello, Jeremy.

Jeremy Goldman:

I will always talk to you, whether you like it or not. Hey, Marcus.

Marcus:

Hello. And also senior analyst covering digital advertising and media based in Virginia, it's Evelyn Mitchell-Wolf.

Evelyn Mitchell-Wolf:

Howdy everyone. Try and stop me from talking about Google.

Marcus:

I know. Okay, today's fact. This is rough. The US has lost at least three nuclear bombs that have never been found.

Evelyn Mitchell-Wolf:

What?

Marcus:

That's right.

Evelyn Mitchell-Wolf:

How do you lose a nuclear bomb?

Marcus:

Yeah, exactly. Exactly.

Jeremy Goldman:

I feel better about not knowing where my keys are right now, so that's not so bad.

Marcus:

2022 BBC article by Zaria Gorvett. So three of them. The Center for Arms Control and Non-Proliferation notes the first one, it's a B-47 bomber dropped an 8,000 pound nuclear bomb into the waters of Tybee Island in Georgia after colliding with an F-86 fighter jet. So there's that one, that was in I think it was '58, '59. Philippine Sea, 1965. A bomber plane pilot and B-43 thermonuclear bomb slipped off the side of a carrier boat, never to be seen again. And 1968, the Thule Air Base in Greenland. There was a cabin fire forced the crew to eject, leaving the plane to crash with its nuclear payload on-board.

Evelyn Mitchell-Wolf:

Okay. So they know where they lost them, they just can't find them.

Marcus:

Yeah. Yeah.

Evelyn Mitchell-Wolf:

Not that they just disappeared out of tracking systems.

Jeremy Goldman:

I like you're putting a positive spin on it.

Evelyn Mitchell-Wolf:

Well, no I'm just trying to-

Marcus:

They didn't just forget where they put it. Like, "Where's that nuclear bomb"?

Evelyn Mitchell-Wolf:

That's what I thought. And so I'm reeling myself back.

Marcus:

[inaudible 00:02:56] "Where'd you put it?" "I don't know. I don't know. I had it a second ago".

Jeremy Goldman:

We know what part of the planet it's on, so we feel a little better.

Marcus:

However, because that is frightfully terrifying and negative, also... This is a new factor of the day for you. According to research by scientists at Northampton University, which is the town next to where I grew up, cows have best friends.

Evelyn Mitchell-Wolf:

I love that. I love cows. I love friendship.

Marcus:

And we're back. And they get stressed when separated.

Evelyn Mitchell-Wolf:

That's precious.

Jeremy Goldman:

Are you milking this fact for all it's worth?

Marcus:

Aww. Yes. And on that note, today's real topic, Google.

All right, we're talking Google today. So let's do it, let's start by talking about how they wrapped up the end of the year. Q4 looked pretty good. And Q4 revenue, this is ad revenue, 2024 was almost identical, revenue growth that is, to 2023 at close to 11%. So 11% ad revenue growth in Q4 of last year... 2023 I should say, and Q4 of 2024. But the real story for me at least was when you zoom out, 2024 growth was nearly double 2023. So what that is telling you is that Q1, 2 and 3 were really good. Q4 growth was similar to the year before that, but Q1, 2 and 3 is where Google really made some significant gains.

We're playing a slice of pie to start the episode off. So Evelyn, do you want to go first? Could you create a pie chart for us as to the reasons, three max, why Google in your opinion was able to double full year ad revenue growth from 2023 to 2024?

Evelyn Mitchell-Wolf:

This one, this is kind of like asking who deserves credit for a class project where one student did the bulk of the work. Then another one had a key idea that really animated the project, but they had less time to work on it. And then the third student actively held the other two back. In this situation, Google Search would be the workhorse, YouTube would be the idea guy, and Network would be the slacker. Network's under performance is no mistake on Google's part, it's a very strategic choice. It's lower margin than Search and YouTube, which are owned and operated by Google. Network attracts outsized regulatory scrutiny for the amount of ad revenues that it contributes.

Marcus:

Mm-hmm.

Evelyn Mitchell-Wolf:

In Q3 2023, YouTube contributed more ad revenues than Network for the first time, and Network has continued to shrink while YouTube has grown. And then over the course of 2024, Network accounted for 11.5% of ad revenues, but since it logged negative growth every single quarter, I'm going to say it gets no pie.

Marcus:

Yeah. It's been negative since the middle of 2022, every single quarter.

Evelyn Mitchell-Wolf:

Yeah.

Marcus:

And now it fell, it was negative 3% for the full year 2024, it was negative 4.5 the year before that. So it did get better, but it's still in the red in terms of growth.

Evelyn Mitchell-Wolf:

Yeah. It's a strategic play on Google's part to sort of remove its dependence on that segment. But as the cash cow that has held its own against rising competition from a variety of angles, I'll give Search 75% of the credit for Google's ad revenue growth. That's roughly the same as its share of 2024 ad revenues. And then YouTube, which drove the most growth but only accounted for 13.7% of ad revenues, I'll say YouTube gets 25% of the pie.

Marcus:

Okay. Jeremy, what would your pie chart look like?

Jeremy Goldman:

For advertising only, right?

Marcus:

Yep.

Evelyn Mitchell-Wolf:

Yes.

Jeremy Goldman:

Yeah, that's what I thought. So yeah, it is really difficult, but I would actually say that... I don't know if I can really bring Network to zero for the simple reason that Network isn't zero, it's just not growing.

Evelyn Mitchell-Wolf:

Yeah.

Marcus:

Right. It's still $30 million a year.

Jeremy Goldman:

Right. It's amazing when we think about this thing that is in decline for a major company and then think about how many companies we cover that are very relevant that Network alone is bigger than, right?

Marcus:

Yeah.

Jeremy Goldman:

I mean that would be a pretty substantial list.

Marcus:

Yeah.

Jeremy Goldman:

I still have to say, and I think probably then for that reason I would give it something like 15. I think that I would actually-

Evelyn Mitchell-Wolf:

So you would give it more contribution for the growth than it contributed to ad revenues?

Jeremy Goldman:

I think that-

Evelyn Mitchell-Wolf:

As it shrunk?

Jeremy Goldman:

As it shrunk, simply because it's... And I know that this might not be as cogent of an argument. I think that from a strategic standpoint, Network is obviously on the decline. However, it could be declining by a far greater rate than it is right now. So I'm not necessarily giving it credit for growing, because I can't because it's not. But at the same time, I think it's worth noting that, again, that is a sizable portion of the overall revenues. I think that, again, if we're talking about YouTube purely from how much it grew year over year, it was 14% up year over year, it's a small percentage of the overall for Google. But at the same time, in terms of the overall cache and in terms of the acquisition of a key coveted demographic and creating really amazing patterns of viewership, particularly with Gen Alpha and other people who are going to be with Google for a long time, I think that YouTube is kind of batting above its average in some ways. So that would be some of my rationale for giving YouTube a bigger portion of the pie, possibly 25 let's say. Network's 15, and then probably the rest, the remainder, the 50 to the core Search business.

Marcus:

Yeah. Yeah. YouTube's performance, you would've thought that it's share of pie in terms of the money it's making would've... I mean it's been growing so fast, going back to 2020, 31%. 44% the year after that. Then it drops, it goes down to 1, but then it goes 8 and 15. But Google advertising itself has been growing at a similar clip, and so the share over that time from before the pandemic to now, the share that YouTube makes up of the ad piece has gone from 11 to 14%. It's not grown as much as I would've expected. However, the growth and the consumer side of things, all the metrics are just doing fantastically well. One of them, YouTube now accounting for 25% of all time spent streaming video on a TV according to Nielsen's gauge. Netflix is at 20% for context. So 25 for YouTube, 20 for Netflix. Put another way, 11% of people's TV time, cable streaming, all of it, is spent watching YouTube.

Evelyn, you said there's a bit of a milestone announced by YouTube as it pertains to TV versus mobile.

Evelyn Mitchell-Wolf:

Oh, yes. Yeah. Neal Mohan just announced that TV now is the primary device from which YouTube watchers watch YouTube. Our EMARKETER's Forecast had that YouTube eclipsing mobile back in 2023, so we've seen this coming really. It's nice to have it confirmed by YouTube, but YouTube is definitely a big player in the TV space and that's part of why I didn't find YouTube's performance surprising. It has a lot going for it. It is a darling of consumers, and YouTube is working really hard to capitalize on those consumer viewership trends.

Marcus:

Mm-hmm. Q4 revenue for YouTube cracks the 10 billion mark for the first time, which is pretty impressive. Let's talk about the Cloud piece of this, because Cloud is kind of a bellwether for how people see Google's AI business investments going. Cloud revenue growth disappointed, but Google's Cloud's operating income was more than double the year over year, over 2 billion in Q4. Evelyn, what do you make of how Cloud is doing and what that says about Google's AI investment?

Evelyn Mitchell-Wolf:

It's a really tangled knot to try and unpick here. I think Google has put a lot of effort into incorporating AI into... I mean to bring it back to advertising, into all of its ads products and into consumer facing products, and that is a really... I think it's harder to look at as a bellwether for how AI is performing for Google because there isn't necessarily a line item on the earnings report that says, "This is how much AI contributed to our ad revenues". But I think Cloud is doing pretty well, all things considered. I think this disappointment was unfortunate considering just the timing of everything going on with DeepSeek and questions around the closed model strategy Google has employed. So I suspect that's part of the market's reaction around Cloud's disappointment. But as far as actual performance, I think it's respectable.

Marcus:

Yeah. That's such an interesting question now, or discussion which is around, Jeremy, how DeepSeek may have reset how we view AI progress. And Jennifer Elias was noting that Google plans to invest 75 billion in capital expenditures in 2025 as it continues to expand on its AI strategy. A month ago that would've been viewed by investors as a good thing, but it feels like DeepSeek has kind of recalibrated the market somewhat. When shareholders read this now they might instead be thinking, "Do we need to spend that much?" Nvidia stock prices still down 20% since DeepSeek showed the world that actually maybe you don't need that much computing power to build advanced AI models. On the other hand, Dan Gallagher of the Wall Street Journal was noting analysts were widely expecting Google to boost its CapEx plans with the company being actually a bit late to the expensive AI party. What do you make of how investors, folks are going to look at these big tech giants paying for AI data centers, paying billions and billions of dollars, in light of this R1 model from DeepSeek saying, "Oh look, we spent $6 million and achieved similar results".

Jeremy Goldman:

Well, a few things there. One thing is Microsoft said that for their fiscal they were going to be investing around that same amount as Google.

Marcus:

Yeah. [inaudible 00:14:16].

Jeremy Goldman:

So we can't put this out of context, right?

Marcus:

Right.

Jeremy Goldman:

Meta said that they were going to put in a little bit less than that, but quite a significant amount. And some of these announcements happened before DeepSeek broke through. People might have overreacted to DeepSeek. I'm not saying they definitely did, but it's just like one of those things where some information comes out, it may or may not be 100% accurate and everybody freaks out and starts doing what somebody else did. It's worth noting that SoftBank, which is not necessarily trying to throw bad money after good or whatever the phrase is. I'm awful at turn of a phrase. Decided to throw in a few more billion dollars into OpenAI, and this was actually after DeepSeek. So I think that the jury is still out on what's the best way to make it big with respect to AI.

But I think the key thing is obviously you have to pay for these investments, no matter what they are. If they're big investments or if they're small and you have to find different ways to monetize these investments. Google's clearly trying to do this both on the Cloud front, both on the consumer facing front from a recommendation standpoint, which they've done forever with YouTube and ad optimization. They've invested in faster ad creation tools to get people to create ads at a faster clip and make that whole entire experience more seamless. So I think ultimately the key thing is are they investing a lot? Yes. Are they finding enough ways to justify these investments? That's something that they're clearly very much focused on this year.

Marcus:

And that's how folks might start to look at this. Nico Grant of the New York Times was saying Google would spend, as we mentioned, 75 billion in CapEx in 2025. That's up, for context, from 52 billion last year or the year before that. So it's a difference of over 20 billion that he notes could have otherwise added to the company's profits. And so I think there is now going to be a bit more scrutiny, even if it wasn't 6 million that they used to develop DeepSeek, their R1 model, even if it was 15 times that you're still looking at 100 million. Even if you times that by 10, whatever it is, it's still a lot, lot less, Evelyn to what you were saying, going with the more of an open source model.

Lots of stories coming out about there's this model that model, there's ChatGPT, there's R1, there's Anthropic, et cetera, et cetera. However, it doesn't seem to have knocked Google off the Search perch, if you will. They still-

Evelyn Mitchell-Wolf:

Oh, Search perch. That's great.

Marcus:

You're welcome. That's why they pay me the medium dollars, Evelyn. And so lots of headlines about how Google might be affected, how people are going to be looking for things online differently, and there's a Capgemini survey that I saw, 6 in 10 people have replaced traditional search engines with Gen AI tools. Now Google has a Gen AI tool, but I'm wondering Evelyn... And it's something we talked about last year, but now we're into a brand new year. How much do you think other Gen AI chatbots, ChatGPTs, folks like that are going to be able to chip away Google's Search dominance this year?

Evelyn Mitchell-Wolf:

Yeah. I think it's important to caveat before I launch into I have a bunch of data to bring to bear on this. But before I get to that, I think it's important to caveat that the Capgemini survey found that nearly 6 in 10 have replaced traditional search engines with Gen AI tools as their go-to for product and service recommendations.

Marcus:

Yes.

Evelyn Mitchell-Wolf:

So there's a specific use case that that applies to. I recently put out a data drop on the state of Gen AI search going in 2025, because there is so much buzz about Gen AI platforms and Google's vulnerability and there's survey data that suggests a critical mass of consumers has decided to defect from Google, and yet it doesn't really show in a lot of different places. Google's Search ad revenues are healthy, we just discussed that. They grew 13.2% year over year in 2024, compared to 7.7% in 2023. So that's positive momentum. Google traffic volume still dwarfs that of its Gen AI first competitors. Similar web data from December shows Google saw 31.8 times more web visits than ChatGPT, which was the next biggest Gen AI search destination. I mean, Google remains the single most critical source of search referral traffic for publishers. Data from BrightEdge shows that despite... Did I take your data, Jeremy?

Jeremy Goldman:

You did. It's a video podcast, people saw me react. Oh well, that's live podcasting.

Marcus:

Throwing things.

Evelyn Mitchell-Wolf:

Well I mean there's so much data about this right now, which is why I put together the piece I did at the start of the year. The data from BrightEdge shows that despite eye-popping increases in referral traffic from ChatGPT and Claude and Perplexity, I mean those growth rates are from such a small base compared to Google that Google's share of referral traffic remained pretty much unscathed at 92.4% in November 2024. So Google is still a giant in this space.

All that said, Gen AI chatbots will chip away at Google search dominance in 2025. Obviously Google is no longer the only game in town. Obviously consumers find the experience appealing, at least for certain use cases. I think advertisers should by no means ignore what's going on with Gen AI search. I just think if we're framing the conversation around what it means for Google and Google's prospects in the search space, Google has this huge, huge advantage over anyone in the Gen AI search space, and that's its ubiquity. Searching anywhere besides Google for any reason takes more time and effort for users right now. And Google also is actively fortifying its positioning against those competitors by adding AI overviews, and this year it also plans to incorporate a Gemini powered chatbot into the search results interface, kind of like where it has news and images at the tab at the top. So it's not going down without a fight.

Marcus:

Mm-hmm.

Evelyn Mitchell-Wolf:

Honestly, I think that Google's biggest weakness is antitrust enforcement. As we all know, Jeremy and I were believe on an episode together [inaudible 00:21:03].

Jeremy Goldman:

Yeah, Evelyn I was actually going to bring up. And Marcus you might recall, but we were talking about this and it's interesting how there are a lot of people who talk about what's going to happen from an antitrust perspective. What is interesting here is that Google seems to be innovating in part as a response to the market, despite the fact that it has such a major market share. So Google's kind of acting not quite afraid, but aware of the fact that they could lose this immense lead. And maybe they could lose it quickly if they don't keep on innovating.

Marcus:

So is Google going to be able to be both things to all people? So for now, I was speaking to someone about this today trying to get a sense of the consumer behavior angle of this. And I was saying to them like, they both use Google and ChatGPT, and I was like, "Well, when will you use each or either?"

Evelyn Mitchell-Wolf:

Mm-hmm. Mm-hmm.

Marcus:

And they were saying, "Well, ChatGPT is for heavier research ideas. Google is for direct answers". And as you mentioned, they're trying to add a Gemini chatbot to Google. Do you think that Google can change that perception of them just being go to them quicker? Because even for shopping now, that's Amazon, people go there. So do you think Google can be the ideas generator, the research place, the direct answers, people all together?

Evelyn Mitchell-Wolf:

I think it can. I think especially if it can maintain its position as the gateway to the internet where it's like a homepage for... I don't have the percentage on hand. I don't know if anyone has the percentage of homepages that Google represents in terms of a data point. That would be a great data point.

Marcus:

I believe they're 90% of searches according to Similarweb.

Evelyn Mitchell-Wolf:

Yeah. Oh yeah. In terms of search volume, really high.

Marcus:

In terms of homepage, yeah. Gosh.

Evelyn Mitchell-Wolf:

Yeah. I mean they are just so... Google is everywhere. And if it maintains that first touch advantage, and it can also sort of show consumers that all of those different kinds of search experiences are at their fingertips with Google, I think it can absolutely maintain its advantage. It does need to improve. I think I notice my own search habits a lot more than I would have if I didn't have this job, and so I pay attention to where I choose to search depending on what I'm looking for. And if I have a question that I know will trigger an AI overview, most of the time I'll still go to Perplexity because I like that layout better. It tends to find things and source them more easily for me to fact check everything. So I use Perplexity, I have it open on my browser pretty much all the time for work. I never use it in my personal life. And all of that, it speaks to a very complex consumer search behavior future. Hopefully we can get some data on that in the next year or two as consumer behavior starts to kind of normalize around all of this embarrassment of choice here.

Jeremy Goldman:

By the way, I think you hit the nail on the head is that there's so much consumer choice right now. And also it's going to keep evolving, because each one of these tools, by the time our next episode when you inevitably ask Evelyn and I to come back to discuss Google, you will see that their feature set will have changed. And all the chief competitors will have changed as well. And who knows, maybe then they'll open up one of those deals let's say between Google and Apple in terms of the de facto search engine that's used on iOS devices. Like if that becomes in play and OpenAI starts to tackle a space like that, then maybe Google's ubiquity gets chipped away at bit by bit.

Marcus:

Yeah, that's a great point. Yeah. Just a question I was thinking to myself, how does OpenAI how do they fit into people's daily lives and how people are going to see them as a tool to help them out with those different tasks? All right folks, that's where we have to leave the conversation unfortunately for today. But thank you so much to my guests for hanging out with me today. Thank you first to Evelyn.

Evelyn Mitchell-Wolf:

Thank you, Marcus. This was super fun.

Marcus:

Yes, indeed. Thank you. Please stop trying to be too shocked. Thank you so much to Jeremy.

Jeremy Goldman:

This was so great, and next time I will bring the data before Evelyn can mention it.

Marcus:

I'll ask you [inaudible 00:25:26]

Evelyn Mitchell-Wolf:

Yeah, next time you'll start with Jeremy, and then I'll... Yeah.

Marcus:

Keep things fair. Thanks to the whole editing crew, Victoria, John, Lance and Danny. Stewart, who runs the team, and Sophie, who does our social media. Thanks to everyone for listening to Behind the Numbers, an ERMARKETER video podcast made possible by Zeta Global. Tune in tomorrow to hang out with Sara Lebow and the retail gang as they discuss how LEGO became the world's largest toy maker.