All right. Good morning, everyone. Welcome to day two of the Morgan Stanley 2025 TMT Conference. We are thrilled to have the CFO of Alphabet, Anat Ashkenazi, with us. Thanks for joining us.
Thank you for inviting us.
Happy Tuesday. Happy first Morgan Stanley TMT Conference. Let me do the disclosures. Please note that all important disclosures, including personal holdings disclosures and Morgan Stanley disclosures, appear on the Morgan Stanley public website at www.morganstanley.com/researchdisclosures. They are also available at the registration desk. Some of the statements made today by Alphabet may be considered forward-looking. These statements involve a number of risks and uncertainties that could cause actual results to differ materially. Please refer to Alphabet's Form 10-K and Form 10-Q, including the risk factors discussed in Form 10-K filing. Any forward-looking statements made today by Alphabet are based on assumptions as of today, and Alphabet undertakes no obligation to update them. So you've been in the role now for about half a year.
Yeah.
It's a very complicated business. There's a lot of moving pieces, like you and I were talking about. Maybe just sort of from an external perspective, you're coming from the healthcare industry over the tech industry. What factors of Alphabet sort of excited you most when you thought about kind of coming to be the CFO?
Yeah. So it's interesting, I guess, the question often of, why did you move from a very iconic healthcare company to another company? And I think about it as really an opportunity to serve as CFO of probably two of the most impressive American companies. And I looked at the Alphabet business, and what got me intrigued is the opportunity to be part of what I view will be likely one of the major, most significant transformations in our lifetime with the prevalent use of AI, both today and where I think it's heading. And as I was studying the Alphabet business and was talking to the management team, I saw a few things that I found incredibly impressive and I thought were really exciting for someone to come in and join as the CFO.
First, I believe Alphabet is probably one of the best-positioned companies to drive significant success at scale in AI over a long period of time. And I saw an incredible team of innovators within the company. And that's something that always attracted me, is the opportunity to work with people that always think a few steps ahead of where they can bring innovation, whether it's to consumers or enterprises or creators with YouTube, et cetera. And I thought, we have this company that has what it takes to be successful. So what a great opportunity to come in now and be able to shepherd that change and momentum.
I know you've had a lot of investor meetings. We've spoken a few times. As you sort of encounter these investor questions, what, in your mind, are sort of some of the most recurring misperceptions or underestimations about Alphabet that you're being asked about that you think just are off base?
I don't know, misperception, but I think I'm not sure why many of our long investors see the fact that we have a really impressive and strong stack. Sundar talks about the stack that we have, but really, across the infrastructure that you've heard us, and you heard me, talk about the investments we're making, but robust infrastructure of data centers, TPUs, GPUs, incredible research teams, and when you're sitting on the outside, what you see is the innovation that we're bringing to the marketplace. We, on the inside, can actually see the level of innovation and the capabilities we have with the team and people, and then, obviously, everyone understands the global reach that we have to billions of users, but the Alphabet business is very comprehensive and very large, and obviously, we talk about Search that's very prominent.
You're now starting to see, well, not starting, but you're seeing the growth in some of the other businesses or platforms, such as cloud and YouTube. Both of these combined, we shared on the last call, are now annualizing over $110 billion a year. That's a massive-sized business. Also behind it, the early innovation that we have with what's in the other bets. You now are seeing Waymo, which we've just launched this morning in Austin, expanding Isomorphic and other opportunities that we have for growth. It's that whole set of opportunities we have across the business.
All right. A lot of growth opportunities. So I've asked you about areas that you think are sort of misperceptions. Now I'm going to throw a few perceptions that I get asked about and sort of your response. So on your first conference call at 3Q 2024, you talked about efforts to re-engineer the cost structure, driving more operating efficiency, et cetera. That was 3Q 2024. Coming out of the most recent conference call, there was a perception among a lot of investors that there was somewhat of a backing away from how much further efficiency you can drive. Your reaction to sort of that idea of there being less OpEx efficiency to come?
Let me start with the end, and then I'll give you the longer answers. So first, there is no change. I think people may have interpreted that change in tone, which was not a change in tone. But on my first earnings call, which was the Q3 call, I set out, I laid out the priorities, which I thought was important for me to describe what the priorities are, the things that I'll be focusing on. And one of the things that I've mentioned was cost efficiencies. It is by no means the only thing. Obviously, you want to drive growth and innovation and adoption of products, appropriate allocation, as well as cost efficiency. We're doing all of that. And once I stated that, my view is now I'm executing. And the entire organization is moving and executing at speed based on the priorities we've laid out.
So I didn't mention it again on the second call. And I did mention the fact that we're going to be faced with headwinds associated with a depreciation that's coming off the significant CapEx investments we're making. So if you think about last year, we had $15 billion in depreciation, which was an increase from the year before. And this year, we'll have more. So with every efficiency you're driving through our operating expenses, you have to offset that depreciation. The efforts are not stopping, and the efforts are continuing. And I think of it as, even for a company that feels that they're incredibly efficient and super disciplined on investments and productivity, there's always an opportunity to push a little further. And it's almost like innovation, right? You never stop.
I like to move away from sometimes people think about productivity or efficiency as an intervention, because it's episodic. You come in and you fix something. And while you certainly want to do that, that's not how I operate. I think of it more as continuous. You never stop. And you do something now, and then the next day you ask, "ok, now we're doing 20% better. Let's get to 30% better," right? And then push even further and even further. And certainly, the more efficient you become, the more difficult it is. But it doesn't mean you stop. I always say it's like a world-class athlete. You don't stop training. You always look for how you can do a little better. And we're looking across the board, whether it's. I talked about headcount growth.
We will be increasing or investing in headcount in certain areas, primarily in AI and in certain areas within cloud, because they're driving growth, but it doesn't mean that it costs the rest of the business we're not prioritizing, so prioritization is critical. In other words, it was already in the media that we've taken some actions in certain parts of the business with voluntary exit programs, and that's just one of these examples of we said this is a priority. Now we're executing. It doesn't mean we're broadcasting it, but that doesn't stop, so people is one thing, and we need to make sure we invest and retain the best, and as I said, having the best researchers, the best minds in the industry is critical. You look at simple things such as our physical footprint, where we are located, the office structure, et cetera.
All that needs to be looked at. How we contract with third parties, how much do we pay for everything, just making sure everything is evaluated very closely. And then we push it forward. And then those efficiencies are what's going to be able to—we can then bring that back into reinvest that into the business to drive that continuous investment in innovation. So not stopping one of the priorities along a few others, which I'm happy to talk about, but certainly one of them. I think probably the bigger focus area is how do we drive growth and how do we get to market even faster.
Let's go there. That's another perception. You're entering the company at a fascinating time of rapid innovation across generative AI, GPU-enabled machine learning, and just changes across the tech ecosystem. There is a perception by some investors who say Alphabet is still moving too slowly. They've built great technologies. They're similar to maybe Bell Labs or a great research lab, but their go-to-market is too slow. What is your reaction to that? And how do you think about sort of ensuring that you're pushing out next-generation consumer-facing products fast enough to stay at your position at the top of the funnel?
So certainly, having tremendous innovation engines is insufficient if you're not getting it to users quickly and you're not driving adoption. So that whole continuum and that flywheel needs to work well. And I think you've seen, if you've just looked at the announcements we made over the past three months, literally every month, sometimes every week in certain months, we've announced something, whether it's Gemini 2.0 or some of our video models or agentic or in other parts of the business, Waymo, Willow, keep announcing new innovation. And that's the other piece of my focus is how do we make sure we get that to consumers, enterprises, creators faster and then drive rapid adoption? We are already in billions of people's homes, TVs, mobile devices, computers, et cetera. How do we make sure we bring that innovation to them?
Now, we do this in a way that is responsible, because that's who we are, and we want to make sure we do that well, but it doesn't mean we shouldn't be testing things earlier, so that's another area of focus that I have, to make sure that we move that rapidly and get that to the market quickly. It is a competitive marketplace, but beyond the fact that it's a competitive marketplace, we have tremendous innovation. I see it. Some of the things that I see, I like to get my hands into things, and some of the innovation that I'm seeing they're showing me in the labs, et cetera, it's just phenomenal. You never thought things could be possible, and I want to make sure our consumers, all of us, have access to it, because it is transformational.
Again, this is one of the reasons I came here, because I do think it is transformational for really across the globe.
Yep. It brings up that it's actually a great segue to sort of the question about pushing from the lab into broader-based products to all the consumers in this room. And sort of the gating factors are what you consider internally for how quickly you're going to push big change to Search. There is this discussion about the Innovator's Dilemma. Alphabet won't change Search fast enough because they don't want to disrupt themselves. How do you think about that internally when you see these great things in the lab and you say, "Wow, this would be great to consumers"? When is it ready to go to consumers?
So I always talk about self-disruption. This is just my—I like to talk about this. And I think you should always look to disrupt your own innovation and not wait for someone else to do it. And that's the thinking that we have. And certainly, for Search, this is probably one of the places where you've seen AI implemented in a really meaningful way. And think about how we all used Search five years ago, even. You're not searching the same way today. First, you're not asking the same questions. If in the past, you'd ask very simple questions, very focused, now you know that you can get a lot more from Google Search. And you're asking a very complex set of complex, much longer questions. We brought in a different way of searching. So now you can search with Google Lens or Circle to Search.
We have AI Overviews, and we're seeing that when we bring that innovation to consumers, we see an increased level of engagement. So we are seeing an increased level of engagement with AI Overviews. We're seeing an increase in commercial queries. So the search metrics are very healthy, but we're continuing to innovate behind it. Sundar mentioned that 2025 is a big change for Search, which means more innovation is coming and more innovation is coming. So we're not going to wait. We're going to bring it to consumers so that they can benefit from it, but I do think we have been changing and will be changing Search and how people search and what they search for. And now they're all, for example, the AI Overviews are powered by Gemini. So we have the power of our AI models behind Search as well.
Yeah, Sundar has talked about a lot of change to come to Search throughout 2025. To your point, I know you shared some new stats in a blog post last night around overall search behavior. So maybe just a couple of questions on it. So can you give us an update on any numbers of what you're seeing in changes from Search from some of these new generative AI capabilities? And where are you in sort of the monetization of overviews and some of these new use cases versus core search?
Yeah. So we shared yesterday that we have over 5 trillion, more than 5 trillion queries a year, which is a really impressive number. I mean, it'd be divided by the number of people, et cetera, per day. And it's not surprising, I think, for all of us. That's how we look for information now. And certainly, the introduction of different methods of searching and AI Overviews have helped with that. And what we're seeing with AI Overviews is when someone is served with an AI Overview, and I'll mention in a second, not every search ends up with an AI Overview. They do get more information. And they tend to then come back for more complex and longer queries.
We're now, and I think we've shared this before, we're monetizing Search with AI Overviews at the same rate as non-AI Overview search, which is great from a monetization perspective. We're seeing all that progress and bringing opportunities for us to monetize even further as we think about where consumers are. Certainly, advertisers as well, as they have more targeted specific searches. You search and you get an AI overview, or it's more targeted now in what you're going to be served with in terms of advertisement. It's beneficial for our advertisers, beneficial for our users, beneficial for us as well.
Great. Now, let me ask you more about the Gemini app. Now, you mentioned you have a big funnel of users who are searching every day on Google, billions of people who are searching. Now, it seems like there's been a change where there's going to be the separate Gemini app now. So walk us through sort of why move Gemini into a separate app? How do you think about monetizing the separate app versus the core search app or Chrome or your other multi-billion user services?
So Gemini now has both a free tier for anyone who wants to use it, as well as a paid tier for someone who wants something more advanced. So there is certainly a monetization option there as well. And we're learning as we're getting more users to use the platform. And it's now even more advanced. And people can interact with Gemini in a different way. It can look through your calendar or Google Maps where you are. If you have an Android or a Pixel, it even interacts in a more advanced way. It can see your screen. So it literally is an assistant now for you. And if you're like me, you have Gemini planning your vacations, detailed itinerary. So I do think there are opportunities there.
The interesting thing is also the dynamic between Gemini and Search, because oftentimes, or sometimes you go into Gemini and you look for, you create something, and you look for information that is not a traditional search, but then you'll use that to go back to Google Search and look for something else. So I think we have multiple options there. Our approach had always been Google for two plus decades is to make sure we build the best products for consumers. And when users use them more, we see a greater level of engagement. We learn more. We then bring forward opportunities for monetization, which AI Overview is a great example of that.
Yeah, the free versus paid subscription offerings. You're not new to those. A lot of times, Wall Street can be impatient with seeing return. They want to see return from all the investment. I think there's an argument that as long as these next-generation agentic capabilities are behind paid walls, they might not be that large. So how do you think about internally of pushing out some of these next-generation agentic capabilities for the free tier, the travel product you mentioned, as opposed to keeping it behind the paywall? What are the gating factors there?
Different variables. Again, we've done it before where we look at what's the user engagement, what are we learning from users, what else are we bringing to consumers, and then we need to make a decision. We make a decision. Obviously, we don't announce it before we make it. But then we can make a decision of shifting this to some place else.
Yeah, ok. Let's talk about CapEx, sort of support, all the AI investments and the ROIC. So we're talking about $75 billion of CapEx this year. How internally are you viewing the company? You want to spend $75 billion. What types of ROIC metrics are you running internally? Or how do you determine the right amount of CapEx, given where we are in this generative AI build-out phase?
So as you can imagine, this is an area I spend quite a lot of time on. And it starts with a very robust kind of governance process by which we look at the demands or the needs across the business, whether it's Google DeepMind or Cloud or Search or YouTube. And we aggregate those needs and look at those needs over the long term, because these investments are made in such a way that we want to make sure that they're used for years. And I'll talk about that here in a second. And we take those inputs in, and we then look at what the needs would be. And we do have the benefit of having TPUs and GPUs when we make those capital investments. TPUs really power our internal efforts. And then for cloud customers, they have options of TPUs and GPUs.
The $75 billion is not all of our technical infrastructure, because we have facilities, et cetera, buildings, but mostly. You can imagine data centers, servers, networking equipment, with servers being the largest component of that. To look at the needs, we then look at what options we have. There are long lead times for this. It's not a light switch. You buy something and you install it, and it immediately connects. It takes time to do these things, certainly if you build data centers, so when I look at the ROI, I do see a continuum of things that will have a return on investment in a very, very near future that are fairly certain, such as cloud customers versus things where you're still building, as you said, kind of the future of AI and everything in between, such as Search and YouTube.
So as we look at our $75 billion, and most of it, as I said, goes towards servers and data centers. And I look at where do we allocate that compute. So in 2025, we expect about half that AI compute to go towards the cloud business. So when you think about it from a return perspective, it's a different profile of return on investment and certainly a different risk profile. Because I think I've mentioned on the call, we exited 2024 with more customers than we had capacity in cloud. So it's not that we're building something where we don't have demand. We already have the demand. Now, again, it will take some time to get this up and running and operational. But we look at this. And the other question is, is fungibility over time?
We have the benefit of having a broad business across different surfaces and different business lines. So I look at if we use this here today, would I be able over time to use it somewhere else so that I can. These are expensive servers, obviously, that we're putting in large data centers. I want to make sure that, one, they're highly utilized, and two, that they'll be utilized over a long period of time. Then going back to your initial question on efficiency and productivity, certainly in $75 billion, I want to make sure that we are investing in the most efficient and productive way, what we bring things in, how we build the data centers. Our strategy is to lean mostly on our own built data centers, which means they're more customized to our needs. Our TPUs are customized for our workloads and our needs.
It does allow us to be more efficient and productive with that investment and spend. And that's incredibly important when you're talking about these sums of money. It's a lot of money. So I look at the efficiency. How do we make sure that even when we build and plan for data centers and we have servers that are going to come in, you want to make sure those are aligned so that when you have a data center ready, then you can put the servers in there, that nothing is waiting, there's no kind of wasted time, and that it very quickly becomes operational live and that we can use that capacity? So it's all the way from scrutinizing the demand, ensuring we invest in the most efficient, productive way, and then allocating in a way that both contributes to growth in the near term and long term.
That's good color. Maybe let me ask one about the GCP and the cloud growth then, because it did decelerate most recently. Was that deceleration driven by capacity constraints? Is the first question. Then as you think about getting the infrastructure and the capacity in the right place, what are the key execution areas to re-accelerate? Do you have to build the products out? Are the products built? You just need the chips to ship them out? Where are you in sort of the driving the acceleration fundamentally?
Yeah, cloud growth in the fourth quarter was 30%, which is lower than 35%. But 30% is really, really impressive growth, and certainly on the cloud base. So it's very exciting to see that growth. Two dynamics that impacted that growth rate in the fourth quarter. One is we lapped a very strong AI deployment in the base period in Q4 of 2023. So it just impacts that year-over-year comparison. But the second is what you've highlighted, which is we had a very strong demand from customers, and we didn't have the capacity to match that demand. In fact, I looked at our list of customers because I wanted to make sure it's real and who we're going to serve when, long list of customers that we're going to be able to get to. But we ended the year with more demand than supply.
And it's really evidence of the great work that the cloud team is doing of providing outstanding products and services for our customers, because we're seeing, and this is an important thing, growth both within our existing customer base and by adding new customers. And I think that's always an important kind of metric to look at internally, because it provides insight into the resilient growth profile for the business. And when you see growth within the existing customer base, you know that they're getting what they need. And it's exciting when we hear from customers that they, on their side, are able to serve their customers in a better way, whether they're able to drive more productivity or efficiency for their customers or offer new services and products that they've never been able to do before.
The partnership with Salesforce is a great example of that, where they're now using Agentforce for their enterprise customers, a great partnership. So the more our customers can do that, the more growth we have in the business. So it was the entire industry's capacity constraint at this point. We are investing appropriately, and we're working hard to make sure we're bringing more capacity online. And we're increasing, obviously, this year the CapEx. Again, it's not an overnight thing. You still need to build data centers, and you need to make sure you put the servers in them, and then you can use them. But it does take time. But I'm pleased with where the cloud business is going. And you've asked about products and services.
Certainly, you see the incorporation of AI and Gemini into our products and services that Cloud is offering, whether we're offering even the TPUs and GPUs options for customers who have both, or the models that we have, which we believe are the most advanced models for developers to use. They have options as well as security, which is really important for, if you think about enterprises, and I think we're leading in the space of security and offering to our enterprise users, and we've recently announced that now every one of our Workspace users have now access to Gemini as well, so they're now able to use that additionally, so a lot of great work is done within our Cloud business.
Got to get through the capacity constraints because the data center is built. Let's talk about cloud profitability then with all that, because there's so many moving pieces of your cloud margins that we think are still below your peers. And so there is a scale opportunity there. But then you do have all the GenAI coming in. And there's a discussion about sort of the incremental margins on some of these GenAI products. So how should we think about the cloud segment profitability over the course of 2025 and into 2026 as AI grows in the mix?
Certainly, we saw in 2024 an expansion in the cloud margins. I will state two things. One is scale, as you mentioned. Scale helps drive more profitability. The second is, and I'll give credit to Thomas Kurian, the cloud leader and CEO, of really driving efficiency and productivity across the business, whether it's where are we hiring, how we run the business, just looking across the business to make sure we do it in the most efficient, really focused way. That is not stopping. We have to continue it, scale as well. As I mentioned earlier, as we're investing more, obviously, there's more depreciation. We work to manage that as well. Just same as we're doing for the entire organization is to work on productivity and efficiency.
Ok. One of the other core platforms, again, it's another billion-plus user platform, is YouTube. You've talked about some of the benefits from generative AI in YouTube. GenAI-powered video campaigns had 17% higher return on ad spend than the manual campaigns. You have other creative tools and diffusion models you've rolled out. So as you bring a fresh lens to YouTube, what are sort of, in your mind, one or two of the biggest opportunities for GenAI to drive faster structural YouTube growth going forward?
YouTube is one of the, it's an excellent example, actually, of the use of AI within Alphabet, and we're already seeing that, so the integration, whether it's for creators, which that's where we start at YouTube, our viewers, or advertisers, so for creators, they now have new tools. If you look at a YouTube Short or a YouTube Long-form video today, it might look very different than what it looked like a year ago, because they have a completely new set of tools, whether it's music in the background that they can use for their creation of production or VO or new video models, so they can really bring new ways of creating content, and the content is just incredibly impressive, and we're seeing that on TV, so it's interesting that YouTube is now becoming the new TV.
It's the number one streamer in the U.S., has been such for, in terms of watch time, for two years now, and you're seeing that increase in TV. People now are sitting in front of, in their living room and watching YouTube, and part of it is the fact that creators are able to create much higher quality content and deliver it to viewers, so when viewers are getting that, obviously, they're watching more relevant content so that gets to the viewers now in AI. We're able to then serve you as a viewer more targeted, relevant content that you're interested in based on different preferences, and when viewers watch and creators create, advertisers come as well, and they have tools as well, whether it's more dynamic ad creation or even getting more targeted, better ads. They're monetizing well as well.
So you do see that across kind of the three pillars within YouTube across creators, viewers, and advertisers being leveraged very appropriately. YouTube is doing incredibly well. We're seeing the growth and the watch time. Shorts, YouTube TV and Shorts monetization, I think we shared before, we're closing that gap on monetization. And I expect we've increased that 30% last year. I expect we'll continue to progress in 2025 as we're bringing more offerings to creators. And then viewers are benefiting from that as well.
The multiscreen YouTube TV is amazing innovation, very underappreciated product innovation, four screens at once.
The interesting thing is we all, if you're like me, you're watching multiple screens at the same time. If you're like my kids, you're screening YouTube on the TV, and you're sitting with your device, and you have your laptop on the side.
Eight things at once.
That's right.
And writing a note. Let's talk about capital allocation. It's interesting. You come into a company that has a very healthy balance sheet and throws off a pretty healthy amount of free cash flow. So what is your perspective on capital returns, the "appropriate amount of share count shrink"? How are you sort of balancing that philosophically?
I look at capital allocation priorities. I start with how do I make sure that we fund the business appropriately, whether it's the capital expenditures we've shared for technical infrastructure or anything else across the business. That has to be the first place where we invest our cash and cash flow, because that is what's going to create a resilient growth profile. That's something I'm focused on, is how do you have a resilient growth profile that's not just for the next quarter or for the next year, but that can actually sustain for multiple years. Couple that with M&A. If there are interesting opportunities in areas that we are focused on, certainly, we would look at those opportunities. The remaining would be returning cash to shareholders, whether it's through share repurchases or the dividend.
We did $70 billion last year in return to shareholders. That's a really impressive amount. But that's how I think about priorities of capital allocation.
OK. Let me ask you a couple on other bets. And I know you talked about Waymo and sort of the launch today in Austin. We think Atlanta is coming later this summer. So how do you think about, so first on Waymo, how do you sort of think about the appropriate amount of investment levels at Waymo the next couple of years to make sure you stay at the knife's edge of autonomous driving?
So if you look at the investments we've made at the other bets the last couple of years, it may seem like we're fairly flat. So we haven't invested more or less, but kind of below the waterline, what's actually happening is an increase in investments that goes towards Waymo. And I've mentioned on the first call, my first call, I mentioned the notion of pivoting. And it's pivoting away from and towards something. And that comes to capital investment, so making choices. And I think growth is about making choices and strategies about making choices. So we're making choices within the other bets as well. And we see an opportunity that goes back to, by the way, the comment on how do you get to scale quickly. That's Waymo's was one of my areas of that's where we need to invest.
We have a tremendous opportunity here, phenomenal autonomous vehicles that can be used in a variety of ways. So scaling that up rapidly, getting to more markets, getting more vehicles is critical. So it looks like we're investing the same level every year, but we're investing more in Waymo. And then we're pivoting from some other things. We've made an announcement about Verily. So we're making choices even within the other bets. I do think Waymo is a really, really interesting opportunity. Hopefully, you've all had a chance. You're all in San Francisco to take a ride in a Waymo. It could be the future of transportation. But beyond that, there are other interesting opportunities within the bets. You asked earlier, what are the things that people don't see? There is such great focus on Search and YouTube and Cloud, rightfully so. But there's Waymo. There is Isomorphic, right?
We didn't talk about that.
This isn't your background yet.
That's my background.
Talk to us about that. Yeah, talk to us about some of the non-Waymo other bets that I think are particularly interesting from your perspective and your background.
So I think Isomorphic is an interesting one. Obviously, Isomorphic has two large partnerships, one with Lilly, one with Novartis, just expanded one of them as well. Multiple targets could be interesting. I always am a little bit cautious when we do things early stage. You want to see them progress. But if it works, it can be really meaningful. We have the Wing opportunity where we're expanding as well. So there are several interesting things. We invest appropriately. But you want to make sure that with these investments, you have certain milestones. And when you see the milestone, that you can advance further with investments. But a lot of interesting opportunities there.
Great. Well, Anat, thank you so much for joining us.
Thank you.
We're excited for all the Search changes in 2025 and everything else to come.
Thanks.
Thank you much.