Meta Platforms, Inc. (META) Earnings Call Transcript & Summary

All right. Good afternoon. Good morning, everyone. We're thrilled today to have Susan Li and Javier Olivan with us from Meta to talk through everything that's going on in the advertising markets and all the excitement going on at Meta. So thank you so much for joining us.

Thanks for having us.

Thanks for having us. We're really excited to be here.

It's great to see you both. Let me start with 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're also available at the registration desk. Some of the statements made today by Meta may be considered forward-looking. These statements involve a number of risks and uncertainties that could cause actual results to differ materially. Any forward-looking statements made today by the company are based on assumptions as of today, and Meta undertakes no obligation to update them. Please refer to Meta's Form 10-K filed with the SEC for a discussion of the risk factors that may impact actual results.

Okay. There's a lot to talk about. Javi, let's sort of -- let's start with you. You've been at Meta for 15 years now?

Yes.

Taking over the COO role really in the last 12 months. There's been a lot of things that have happened at Meta over that period. Maybe just talk to us, as you sit here now and look into 2023, what are the key priorities that you're most focused on for the platform?

Yes. Thanks, Brian. Thanks for having us. We were talking about taking on roles at challenging times and sports analogies. And I was talking with a friend that is a big New Zealander guy who used to play professional football, American football. And he said, there's three types of passes. There's the short pass, the long pass and the hospital pass. I don't know which one we're on in this case. But my focus really has been on efficiency and how to drive the business and the core business. And when you compound growth and resources over 15 years nonstop, there's a lot of opportunities to do things in a more efficient way. So this is something I started doing regardless, but it obviously became critical, given the current environment. And as Mark announced, this is a year of efficiency. So on my side of the house and what we're really looking into is how do we accelerate revenue? How do we make more efficient use of the capacity, the infrastructure? And how do we continue executing on the core things we need to, whether it's like integrity, all the businesses and products and services that underlie the Family of Apps, okay?

There's a lot there I want to dig into. Maybe let's just sort of level set from a macro perspective, Susan, a little bit. Any updates on sort of what you're seeing in macro advertising demand? Any pockets of strength, pockets of weakness, sort of key advertiser trends you would highlight to start the year?

Yes. Well, first of all, again, thank you for having us. I hope someone out here is keeping count of the number of times the word efficiency gets said in the next 30 minutes. We'll do our best. So to your question, so there's a -- we try not to say too much about the quarter in the quarter but here's what I can say. So when we look back, 2022 is a really challenging year for the business. We certainly experienced the overhang from macro in Q4, and that was also part of what we saw coming into Q1. And when we look back at Q4, we saw financial services, technology, those verticals were kind of the largest negative contributors to the year-over-year revenue decline we saw in Q4. But we also saw that some of our largest verticals, e-commerce and retail, those remain challenged. Now the trends are relatively less challenged than they were in Q3, but nonetheless, sort of lots of volatility in some of the biggest verticals for us traditionally. There were some pockets of resilience too. Health care and travel were sort of year-over-year contributors to growth, but again, those are just much smaller for us. So I think on the whole, it's still a choppier demand landscape and that really informed our Q1 revenue outlook that we gave on the call. And you asked about advertiser sentiment. And so I think that's been a little mixed depending on the region. In Europe, where they're certainly still experiencing issues related to the war in Ukraine, related to inflation, related to energy, sentiment is less positive. In Lat Am and Asia Pac, it's more so, and I would say North America is somewhere in between. So it's a little bit of a mixed bag depending on where in the world you are. But on the whole, I think advertisers everywhere certainly see us as an important part of their business strategy and as part of their efforts to reaccelerate growth going into 2023. And when I think about sort of our positioning in 2023, I feel like across both supply and demand factors, on the supply side, we're pleased with usage trends. We saw all-time highs in DAUs across Facebook, Instagram, WhatsApp in Q4. Reels is scaling well. We announced on the Q4 call that Reels plays had doubled over the last 12 months. Reels reshares, which are kind of a signal as to how the social flywheel is doing, had doubled over the 6 months prior. So engagement trends, I think, are looking healthy. And then on the demand side, a lot of our sort of back-end monetization work, it's not sort of the most glamorous stuff but it's like continuing to make the ranking and recommendation models incrementally better over time, paired with our AI investments, I think we've seen meaningful gains there. And so we feel like we are well positioned, I think, both to navigate the current cycle and then certainly to benefit if and when the macro landscape improves. So I think we're feeling pretty good about how we're going to be able to navigate this.

Great. There's a lot of AI and machine learning been doing in the background for a long time. I want to get into that but add another efficiency to the tally. Maybe Javi, the discussion about the year of efficiency, it almost did come across externally as there's almost a cultural change at Meta to really focused on just durably running the company more efficiently, not just a single year but going forward forever. So maybe talk to -- I know it's so early, it's only March. But talk to us about where have you made the most progress on improving efficiency. And as you sort of look for the rest of the year, what are still the areas where you say we can really run this business a lot more efficiently?

Yes. So like with every project we tackle on, we like to think in terms of like understand, identify and execute. And there's different projects running in parallel and there are different stages. So there are easy things that you find when you start tackling these problems. So for example, we have a handful of different marketing teams spread through the organization that makes a lot of sense to consolidate them under one. Now you have one measurement team, one creative team and you just make a much more efficient use of resources. Similar with our integrity teams. We have a business integrity team, a user integrity team, two different operations teams. So that's now all under one leader. And obviously, then you can just move the resources much faster. It just makes you more efficient. It's not all about organization. There's also other things like the use of our infrastructure. And if you look at how we've been really growing over the last 15 years, it's been all about how much performance can we get for what because what has been the bottleneck? How much power can we get into the data centers? Now as you shift into driving the business more efficiently, you think in terms of how much performance you can get per dollar. And that automatically trickles into a lot of micro decisions that happen across the teams. And those kind of things tend to take a bit more time as they permeate through the organization. So just another example, product teams, when they develop products and then they obviously use capacity. In the past, it's been like take as much capacity as you need. Now we're trying to give them a very clear view of how much is their product costing. And some of it is easy because there's dedicated resources but some of it is hard because they're shared resources, memory, network, et cetera. So just having the right accounting giving it to the team so they can make the decisions and then ultimately, potentially goal them on those. Those things are not step changes. Those are angle changes that will take them to play -- time to play. But the last thing, there's like a lot of these things sound like cost-cutting and you really don't cut your way into growth. And what we're finding, though, is by focusing on these things, you start finding ways to actually drive incremental revenue in much more efficient ways. And one example is leveraging our marketing and product as opposed to just rely on call centers and humans, and it sounds simple. But if you really integrate those systems together and that's something I'm really focused on, bringing this business side and the product teams together, you can now have the tooling, be connected to what's happening on the ad interfaces. And then the humans only call the advertiser at the right time with the right message and at the right point. And that really helps it make it more efficient and just drive actual more revenue.

Okay. A lot of those changes, it seems like they sort of they take time and they grow on themselves over time. Susan you've laid out the initial 2023 -- the latest 2023 OpEx guide. How should we think about puts and takes around that and potential further source of downward pressure on that OpEx guide if you do become more efficient over the course of the year?

Yes, it's a great question. And we had lowered our total expense guide from the Q3 to Q4 call. And so really, our efficiency work began in 2022 and has been ongoing into 2023. And when we gave the total expense guide on the Q4 call, that reflected, we knew this was a year of efficiency, and reflected all of the cost-saving measures that we had identified up to that point. But that work has certainly been ongoing. First of all, we're continuing to look across the company, across both Family of Apps and Reality Labs and really evaluate, are we deploying our resources towards the highest leverage opportunities? And I think this is going to result in us making some tough decisions to wind down projects in some places, to shift resources away from some teams. You saw us do this already last year when we shut down the Portal device line in the Reality Labs division. And this is really hard. I mean, these are all sort of great products and great ideas, but they aren't necessarily kind of the highest priority place for us to invest now. So I think you'll see that we're going to be in an ongoing reevaluation and reprioritization process. The second thing that we've talked about also is we're really looking at streamlining management layers and streamlining cross-functional sort of teams, processes, reviews. And this sort of not only serves us -- serves to make us more efficient and more productive, but it's also really in service of making us a company that can build more and ship faster and making sure that core tech product development is really at the center of what we do. So when we look at all the work that we're doing right now, I expect some of these efforts will translate into further cost savings, and we will update investors as we make those decisions. At the same time, I want to flag that the thing we're really focused on is the long-run cost structure. And so in that sense, in some of these places actually, getting to a better long-run cost structure might mean incurring a little more cost upfront. In both some of the core product development areas as well as a lot of our business processes, these are places that there are bodies of work which could really benefit from heavier automation, but we're going to need to invest upfront to build the tooling, to build the infrastructure that will enable us to do so. Now again, I do think all of this is going to sort of ultimately help us be a more profitable, more financially resilient company going forward. When I look ahead, we talked about the macro landscape. And although I feel good that we're doing everything we can to be as well positioned as we can, I mean, it's -- there's a lot of uncertainty out there, both from a macro landscape but also from a regulatory perspective. And so I want us to be in a place as a company where we can navigate any number of possible challenges, especially given sort of the magnitude of our long-run ambitions in both AI and in the metaverse. And the other thing I would add is, I mean, profitability is near and dear to my heart, but this is not all in service just of cost-cutting, right? I mean, a lot of the things we're talking about, streamlining management layers, reducing cross-functional reviews, this is also in service of making sure that Meta is, frankly, a more productive, more enjoyable, more rewarding place to work. We want people to come to Meta to do the best work of their careers. I mean, I came to Meta to do the best work of my career, and I've been hopefully doing that for 15 years. So we're trying to make this the best place to work and we actually think a lot of these efforts are in service of that also.

That's great. That's great. And one of the big areas you talked about, you have a lot of people doing their best work is around AI. It's not new to you. I think it's -- my question is sort of you, sort of remind everyone listening in the audience of ways in which you've been using AI to both drive engagement and revenue to date? Any math or quantification on the AI benefits you can share? And then as you sort of look at the platform, what are the areas you really see AI further charging more revenue or engagement?

Yes. I mean, I can say, I can see why there's a lot of focus these days on generic AI. I just saw some coming out of state so obviously, a lot of excitement for the right reasons. One of the things we announced last week, I believe Mark talked about this, is we had a bunch of different teams looking at this more from a research point of view, spread through the organization and now they are like centralizing one team. So that's going to really help us turbocharge our efforts on the front. And the applications there are, of course, very natural to our business. Large language models, combined with our messaging apps, have a very natural fit. If you think about generation of images, video, they can help creators by giving them the tools or filters. They can have advertisers to do creative content generation and testing multiple creatives at once. But as you were saying, machine learning and AI have been core to our business for many years. And if you look at how they help us drive engagement and you look at the progress we've made in our recommendation engines, for example, with Reels where in the last year, we've managed to double the number of plays across Instagram and Facebook. And this is based on the new models running on GPUs. That's what allows us to do this kind of progress. And similarly, on the core business and ads, how do we make more relevant ads? The same, it's machine learning algorithms, AI that have allows us to make, so that we can drive, relative to Q4, and you look at the year before, 20% more conversions and more to the advertisers. And we managed to do that and we know all these improvements because we keep hold out. That's what's driving our ability to drive more conversions for advertisers. And we're also doing at a lower cost per action, which means we're effectively driving more value. But yes, these are long-term things we've been investing for years, and this is just a natural evolution of this model.

And part of that has been working to overcome some of the ATT and mobile attribution challenges. So can you help us sort of understand where you are now and sort of that path to close the ATT measurement gaps? And which tools or new products you think have had the biggest impact in closing some of those challenges?

Yes. So the strategy there really has remained the same for a while, and it's really focused on two fundamental things: one, making the ads more relevant and perform better; and two, bring the conversions on site. That's really been the path of attack there. So on the first one, using AI/machine learning, we were talking about some of the improvements we've been driving on the ads relevance. But machine learning and AI also helps you with things like automation of campaigns. And this is bread and butter, but it allows you things like Advantage+ shopping plus to test 150 different combinations of targeting, ranking, creative. And the system optimizes it for you so you just don't have to be manually tweaking the campaign all the time. Similarly with measurement, measurement is critical for advertisers. They make decisions based on what the campaigns are doing. So more powerful machine learning and AI is allowing us to increase the attribution window for campaigns from 7 days to 28 days and also give some cuts like on gender, age that were not possible before and are now, thanks to the improvement on the algorithms. On the other side, we're trying to bring the conversions on site. And I particularly like that one because it just drives also a better user experience. In the past, you clicked on the ad and you were going through a website where the context was lost to do some particular thing. And if you think about the funnel of awareness, intention, decision, action and we can streamline it by keeping it more on site, it is a better user experience and it works better for advertisers. So examples there are lead ads. If you think about it, lead ads, lead generation, used to be an ad you click, you go to a website, you go through a flow and then you try to capture the e-mail address or the phone number. Why do all that if you can do it only in line? We know there's always friction when you just swap contest. So that's one example. Click-to-messaging ads, working amazingly well, over $10 billion in run rate. Those are at that point to opening a thread with the merchant, and now the merchant can have a conversation with the consumer and from there, eventually get them to the decision, the action and the sale. And then similarly, we're seeing things on commerce and shop ads, where by using again, machine learning, the algorithms can determine if a merchant has synced their catalog with us, we can automatically create a shop for them inside Instagram, inside Facebook. And now the ad can decide for this particular user at this particular time for this particular product, is it better to send them offsite? Or is it better to send them to the on-site shop because we know it's going to convert better? And by doing this kind of automation, still small-scale, hundreds of millions of dollars of run rate in revenue, yes, but very promising results because it just does better when you can do that level of decision and on-site option.

All the innovation and the GPUs, Susan, that will cost money.

Yes, it does. A lot.

You're running the budgeting process with CapEx. Maybe just talk to us sort of about the -- the puts and takes on '23 CapEx and what you're spending on as we sort of do go down this path of having more AI tools and more advanced tools. How should we think about the long-term capital intensity on a per minute, per user basis, however you think about it?

Yes, great question. We spent a lot of time talking about CapEx. It's been a big focus, shared focus in our budgeting process. And the big components for us are servers and data centers followed by network infrastructure. And for us, the surge in CapEx that we're seeing in 2023 really is around building AI capacity at scale. And we've talked a little bit about how we're playing a little bit of catch-up here in terms of the AI capacity specifically. So the majority of our server spend in 2023 is going towards GPUs. And the new data center architecture that we talked about on the Q4 call is, again, primarily in service of deploying those GPUs. Although one of the benefits of the new architecture is that it's actually flexibly -- it flexibly supports both AI and non-AI use cases so we don't have to make upfront long lead time commitments to two specific capacity use cases and data center types. This sort of one architecture is going to allow us to accommodate both and basically build as we need, which is great. And in terms of our AI investments, right now, the -- basically all of our AI capacity is going towards ads, Feed and Reels. And where we have been able to deploy GPU clusters at scale, which is still relatively early, we're seeing really encouraging results. And so what we can measure, what we feel good about. And our focus here is basically on continuing to measure those returns in a way that informs our future level of investment. The place where I think it's a little bit different is generative AI, which we've been talking about here. And that's just a place where it's much, much earlier on. And to date, our focus has really been on setting up the right teams and talent. And right now, generative AI is not a meaningful driver of CapEx for us. But it's a place where I absolutely could see there being a compelling opportunity to deploy more capacity in the future. And if and when we do that, I think we'll be just so much earlier in the life cycle of that endeavor and so much earlier on the return curve. It will be a little bit hard, I think, to measure in the very near term, the returns on that capacity in the same way that we can in all of these sort of ranking and recommendation areas where all of our AI workloads effectively are going towards today and where we've, I think, developed best-in-class measurement systems. So when I think about the absolute levels of CapEx, what I'd say are it's going to be very ROI-oriented as it pertains to the existing sort of ranking and recommendation work. And then we're going to have to sort of look at new opportunities like generative AI and make sure that we're able to invest there as we need to. And on a relative scale, as we think about capital intensity, which for us, we, so far, have really thought about as CapEx as a percentage of revenue, we expect to bring that down over time. So we recognize this as a place where we really have to balance, I think, our ambitions, to build a lot of AI capacity towards a lot of compelling use cases for us with the need to really, I think, deploy and invest in CapEx very efficiently.

That's good color. The new architecture on the data centers that you mentioned, I wanted to -- it's a jumpoff, whoever wants to take it. I love just any more color around any technological breakthroughs or advances that are sort of enabling these smaller, more modular data centers that you're building. And how does the -- how do the public cloud players sort of play into the forward capital allocation for Meta?

Yes. So I'll take on this one. If you think about the two ends of the spectrum, right, you can have hyper-optimized, completely vertically integrated data centers from the Linux kernel networking, software stacks, all the way to the top applications, hyper-optimized for what you're trying to do. On the other side of the spectrum, you have the public cloud, absolutely multipurpose commodity you can buy or sell off the market. We are definitely closer to this side. And we've been doing this for years with CPUs. Now what happens is we have years and years of experience optimizing for the CPU type of load. As this new technological disruption happens with GPUs, which is a very different beast in terms of how you manage them in the data center, how you service them, how you do the interconnection within rooms in the data center, how much bandwidth you need to interconnect them. It creates a dislocation from what you've been doing so well for so many years. And to be honest, we were a little bit late through the trend so we're playing a little bit of catch-up, which is why Susan has been seeing our CapEx investments going a little bit higher up as a percentage of revenue. But again, with the real goal of once we kind of get this and start optimizing, again, bringing it down. In addition, we can explore along this spectrum of going towards public cloud. Are there other options in between? Because ultimately, you need to be doing forecasting of how much capacity you're going to exactly need. And because you can't be perfect at forecasting, now you're trading off headroom, which costs, How much CapEx do you leave sitting down, not using? But it's helpful because you can ramp up actual usable capacity quicker. So there's a curve of CapEx and a curve of time to usable capacity that we're kind of trying to optimize.

Okay. Let's talk about the user behavior a little bit on the platform and the emergence of Reels, where you talked about the doubling engagement. Over the years, the use case of Meta has continued to expand, whether it's friends, communities, sources of news and now you have Reels and short-form video. So any update there on what you're seeing for Reels traction, the extent to which it's proving to be incremental time spent? And how do you think about the long-term monetization of the platform, Reels versus Feed or Stories?

Yes. So I talked a little bit about some of the most encouraging Reels engagement trends earlier, and I think Javi alluded to them, too. And so Reels, I think, has really been, I think, one of our most exciting areas of engagement growth, not only because I think we've, I think, done a really good job building for this particular format and the general shift towards short-form video. But also, I think for us, Reels was the entryway into kind of this broader ecosystem of what we call in-feed recommendations and bringing unconnected content, I think, more broadly in front of people and that's been contributing to engagement also. And so I think it's always, I think, encouraging when you see your core product strategy sort of translate into engagement growth in the way that you expect and hope. So on the monetization side, there are really three areas of focus for us. I described them sort of as work on demand, on performance and supply. On demand, what this is really about is basically unlocking the enormous pool of advertiser liquidity across the platform for Reels. And so we've done a lot of work basically enabling the vast majority of our ad objectives to be available for Reels, and there's still a few more formats that we have to sort of unlock and so we're working on that. But I think this is a place where we've made a lot of progress, and we're seeing that 40% of our advertisers across Facebook and Instagram use Reels, and we're just doing the remaining work that enables advertisers to extend all of their campaigns on to Reels, should they choose to do so. On the performance side, this is where we're really leveraging the machine learning work to enable us to basically test very quickly at scale lots of different formats, templates, things like where should a call to action button on an ad go? How big should it be, et cetera, to make the ads basically as performant as possible? So that's a place where we're investing a lot now. And on the supply side, Reels is still sort of earlier in its ad load journey than more mature surfaces like Stories and Feeds. So there's room for ad load to grow. But there is a structural challenge here for us, which is that people spend more time watching Reels than they do viewing pieces of content and Feed and Stories. And so there's not as much opportunity to introduce ads per time between pieces of content. So it's a place where we're going to need to think over time how to be a little more creative about how we think about ad load, where to insert ads, that sort of thing. So there's more work to do there. In general, it certainly, every time we introduce monetization into a new surface, it takes time for that surface to catch up to surfaces that have matured through years of optimization. And we saw that it took years for Stories to catch up to Feed on Instagram in the United States. And we expect Reels may take also years and probably a little longer than Stories for the structural reason that I mentioned. But at the same time, we're excited about the progress in Reels. And in particular, because Reels is incremental to overall time on the platform, that's what is going to enable us to turn Reels from what is right now an overall revenue headwind into sort of -- to a breakeven place by end of this year, early next, because ultimately, Reels is actually driving incremental time on the platform that we can monetize.

The other source of -- that's great. The other source of incremental revenue has been the click-to-message ads. You mentioned them a little before, Javi, the $10 billion run rate. In the past, we've written about sort of the layer cake of opportunities for you guys in revenue. And this seems to be a budding incremental source of revenue. So can you sort of talk to us about, how have you grown that business? What are the channels that are driving that business from an advertising perspective? Where are those advertisers geographically? And in order to take it from $10 billion to $20 billion, what needs to happen?

Yes. No, that's definitely one of the most fascinating trends. And within the click-to-messaging business, there's obviously click-to-messaging on Messenger, click-to-messaging on WhatsApp. And that on Q3 last year, I believe we announced, it's growing 80% year-on-year. And I've been a heavy user of WhatsApp since 2010. That's when it started because in Spain, it was one of the countries where it grew the fastest. And as I took on the role last year and I was traveling the world, honestly, I personally was surprised about seeing how people are using it in emerging markets. And it's hard for people here in the United States to understand what is it to live in a country that really communicates on WhatsApp. But even for me, it was hard to understand what is really happening with people communicating with businesses in places like Latin America, Southeast Asia. And there, it's not really just limited to people talking to small and medium businesses to the baker in the corner. It's large enterprises like General Motors in Brazil. They run a campaign and over 60% of their cars that are sold were started in WhatsApp conversations. They sold over 5,000 cars that started in WhatsApp conversations. So that's one thing I would encourage everyone to do is just really intuitively understand what is happening because it really is transformative. And when you think about the convergence of those trends and people communicating that way with businesses and the advent of tools like large language models, that can make that automation so that it's cost effective in the developed world, I think there's a real interesting potential that might be the first time that I see Meta, a business model really starting in emerging markets and then backing back into the developed world. So that's an area that I'm definitely very focused on.

That's great. Well, we can't wait to continue to monitor the year of efficiency with still a lot of innovation. Susan, thank you, and Javi, thank you so much. Thank you both.

Thank you.

Thanks, everyone.