TRANSCRIPT
Q3 2023 — Follow-up call transcript
Q3 2023PDF · 15 pages184 KB
1 Meta Platforms, Inc. (META) Third Quarter 2023 Results Follow Up Call October 25th, 2023 Operator: Good afternoon. My name is Dave and I will be your conference operator today. At this time, I would like to welcome everyone to Meta’s Third Quarter Results Follow Up Q&A call. All lines have been placed on mute to prevent any background noise. To ask a question, please press “1” then the number “4” on your telephone keypad. This call will be recorded. Thank you very much. Ken Dorell, Meta’s Director of Investor Relations, you may begin. Kenneth Dorell: Thank you. Good afternoon and welcome to the Follow Up Q&A call. With me on today’s call are Susan Li, CFO, and Chad Heaton, VP of Finance. Before we get started, I would like to take this opportunity to remind you that our remarks today will include forward looking statements. Actual results may differ materially from those contemplated by these forward looking statements. Factors that could cause these results to differ materially are set forth in today’s press release and in our quarterly report on Form 10-Q filed with the SEC. Any forward looking statements that we make on this call are based on assumptions as of today, and we undertake no obligation to update these statements as a result of new information or future events. During this call we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today’s earnings press release. The press release and an accompanying investor presentation are available on our website at investor.fb.com. And now, I’d like to turn the call back over to Dave for the first question. Operator: Thank you very much, Ken. Your first question comes from the like of Mark Mahaney with Evercore ISI. Your line is open. Mark Mahaney: OK. Can I throw two questions out, please? Susan, on the commentary about kind of the softness to begin the quarter, could you spend just a little bit more time on that? Did it start soft and then kind of quickly or somewhat -- 1 of 15 -- 2 recover, which is kind of what Snap indicated, or does it look like the softness has kind of just stayed the same since October 7th or has it gotten worse? Just anything on that. And then, how do you figure that into the guidance? It’s clear that your guidance range is wider at $3.5 billion than it typically is. So, that’s clear. But does this -- is there something about the ends of the spectrum that you’re trying to signal like quick resolution, which -- whatever -- would mean we’d be at the high end? Just talk about how you figured out that the point range is. And then if you could please, Reels and click-to-message ads, you’ve at times in the last three or four quarters kind of quantified where they are, $10 billion levels. Can you give us any updates on where those absolute ad dollars are for those two areas? Thanks a lot. Susan Li: Hi, thanks, Mark. So on the comments that I made on the call about the softness at the beginning of the quarter related to the geopolitical events in the Middle East, first certainly I should say in the region we’re seeing a decline with Israel based advertisers. That, in and of itself, is not a material part of the business. I said we’ve seen some advertisers pause brand campaigns since the start of the conflict. We’ve also seen a number of those campaigns resume. And more broadly the more performance oriented areas of our business like online commerce have been less impacted. So really, I think what we are trying to do here is be as transparent as possible about what we’re seeing in our own revenue trends, which we obviously monitor very closely. We recognize from prior history that geopolitical events that can occur -- in the opposite side of the world, can have very small measurable sort of first order impact that you can see. In this case, for example, the specific activity related to advertisers in Israel. And at the same time, we often do see a broader demand softness that coincides from a timing perspective with that conflict. -- 2 of 15 -- 3 At the same time, those events are somewhat -- that tends to be bounded, right. These are -- we have certainly seen recovery since the events in Ukraine, and so I’m really -- I’m really trying to signal here that this is just an uncertain period. We have seen some impact. Again, we’ve seen some recovery and we just want to be clear that there is uncertainty related to events that are going on in the world more broadly. And the widening of the range is really just intended to signal that we think there is a wider range of plausible outcomes. Chad Heaton: Yes. So on the second question you asked about Reels and click-to-messaging and about whether we can quantify the dollars, so we don’t quantify -- we haven’t quantified the dollars of those or provided updates on that. But I can give you a little bit of color on both. So starting with Reels, we made steady progress over the past year of bringing down the revenue headwind that continued in Q3 as Reels reached overall revenue neutrality. That Q3 improvement was driven by improved monetization efficiencies through the ongoing work to make Reels ads more interactive and relevant, as well as continued incremental time spent gains. On the CTM side, we saw sustained strong momentum in CTM ads. We saw year-over-year revenue growth was faster than the overall ad revenue growth in Q3. The click-to-messaging -- or excuse, the click-to-Messenger piece is the largest portion since we started there first. On click-to-WhatsApp, that annualized revenue is already in the billions of dollars, and it’s continuing to grow very quickly and in click-to-Instagram Direct is the earliest portion but that is growing quickly as well. Mark Mahaney: OK. Thank you very much. Operator: And your next question comes from the line of Brad Erickson with RBC Capital Markets. Your line is open. Brad Erickson: Great. Thanks. So I guess first you called out China, I think, to the driver of growth in pretty much almost every region. Can we just get any help with -- 3 of 15 -- 4 understanding how big that is now in terms of percent of revenue or how much it contributed to growth? I think given you’re growing well into the double digits in all regions and calling it out in every region, I would think it has to be at a point now where we can maybe talk about that. Any color there would be great. Susan Li: Thanks, Brad. We have not broken this out or quantified it yet. Obviously, we will do so when we’re at an appropriate time to do so. Like you said, it is contributing to sort of strength across many regions as well as across some of our strongest verticals in online commerce and in gaming. And so, it is a -- it’s definitely a place where we’re benefiting from strong investments, especially from a few of our larger clients. I called out the broader based general China strength, and we’ll share more here when we have more, but it is absolutely a driver of strength for us in Q3. Brad Erickson: And then, Susan, thanks. That’s great. And then maybe just one follow up. Just curious on the Q3 results. Did that China revenue contribution, did that come in ahead of plan in a material way? Susan Li: I don't think we typically comment on our internal goals in this forum. Again, I would just say that it’s been an outsized contributor to the Q3 acceleration, it indexes as well with our -- with verticals that perform particularly well on us. And I think it’s benefited from also a combination of macro factors and geopolitical factors that we referred to on the call also. Brad Erickson: Got it. Thanks. Operator: Your next question comes from the line of Ron Josey with Citi. Ron Josey: Great. Thanks for taking the question. Susan, I wanted to ask maybe a bigger picture question around just the rise of unconnected posts and AI- recommended content. Can you just talk maybe philosophically, like what is the right mix, if there is one, of that recommended content from AI versus from a user’s friend and family? Just we’re seeing so much strength from the content that’s being provided to us, wondering what the right mix is. -- 4 of 15 -- 5 And then, a quick follow up just on your comments around the FTC looking to modify the consent order. Any insights there on timing or how we should think about that? What the potential impacts could be would be helpful. Thank you, Susan. Susan Li: Yes. Thanks, Ron. So on the first question, the way -- we’ve certainly called out the sort of strength in unconnected content and our recommendation engine in terms of driving incremental engagement across both Facebook and Instagram, and Mark called out some of the stats. I don't know that we think there is a sort of optimal mix, and I think actually the optimal mix is going to depend a lot on the person. So, I think there'll be heavy individualization in terms of what is the optimal experience for each person. I am a person who, for example, as part of my experience, I quite enjoy unconnected content. I love seeing ads probably more than anyone, and I think my father when he uses Facebook would like to see nothing but photos of his grandchildren. So, people have, I think very different preferences. And our goal, ultimately, is to serve to each individual the optimal mix of content for them. The other thing I’d say, too, is I just think that the sort of presence of AI in different parts of the ecosystem, I think, is going to become -- it’s going to be more important in different ways. So, your question was specific in terms of the ranking and recommendations piece and AI powering those ranking models. A lot of the AI work we’re doing today, too, is about making it easier to generate content, for everyone, including obviously creators who specialize in this but also friends and family. And so, our hope there is that those AI creative tools will both make it easier for everyone to create more engaging and compelling content and that'll increase the volume of content, again, from both creators and friends and family that can be surfaced by our ranking engine. -- 5 of 15 -- 6 And then, your second question was about the FTC. So, the FTC initiated an administrative proceeding to rewrite the privacy order in the spring and, has proposed among other things, imposing significant limitations on our business, including our ability to launch products and modify existing ones. And so, we’ve asked the federal court to block the proceeding. We expect a ruling in the coming weeks. And if we’re unsuccessful, our response in the FTC proceeding is currently due at the end of November. We do intend to contest their proposed action and we don’t know how long it will take it for them to consider our response. So, this is an ongoing litigation, so I’m not planning to really comment further about it. But just to give you a sense of timing. Ron Josey: That’s helpful. Thank you, Susan. Operator: And your next question comes from the line of Ken Gawrelski with Wells Fargo. Ken Gawrelski: Thank you for that. A couple questions if I may, please. First on the CapEx side, I think there was some considerable fear around what a -- how big the CapEx budget or forecast would be for ’24, but could you just talk about with the guided range, could you just talk about the -- any of the constraints that you may have had, either on the supply side -- or from the supply chain side - - or how much of it is variable based on adoption of your AI products out there, relative to just overall budget? And then, I’ll ask … Susan Li: Sure -- oh, okay. Do you want to tee up the second question for us? Ken Gawrelski: Sure, so yes, the second one, please. Sorry about that. On the OpEx side, could you talk -- you gave us the OpEx guide but then you repeatedly said, Susan, about the -- you talked about the hiring backlog commentary and that hiring should be greater in ’24 than what would otherwise be normal. Should we then -- what are we really -- what’s the key message to take from that? Is the message that when we look -- when we think about ’25 and ’26 -- 6 of 15 -- 7 in the outer years that we should have a more normalized kind of staff net additions? And it will be reflected more in the OpEx numbers going forward? Is that -- what -- I’m just trying to get a sense of you’ve made that comment repeatedly. What’s the key takeaway there? Susan Li: Well, thanks Ken for the questions. So, the first question you asked about the CapEx guide. So, I mean, first I would emphasize that we’re certainly continuing to make significant investments in our infrastructure, and we see a lot of compelling opportunities to do so across both the core AI and generative AI work. There are basically two halves to the planning equation. On the demand side, we’re continuing to scrutinize where we are putting additional capacity to ensure that we’re investing in areas with strong returns today, areas like our core AI work where the returns are more immediately measurable, as well as some of more forward-looking strategic priorities like the generative AI and advanced research we’re doing and just finding ways to identify cost- savings opportunities, optimize our workloads, make sure that we’re really streamlining our capacity needs wherever we can. On the supply side, our ability to procure the capacity we need does influence where our ultimate CapEx spend will land in a current period. And we are expecting some supply constraints with GPUs next year, particularly in the first half. We’re also working on different plans to mitigate that, in part by ramping up our adoption of our own customer silicon chips, particularly for the ads recommendations work. But the expected supply constraints, as far -- as best as we can foresee are also factored into our outlook. I would just say this is a very dynamic area of planning generally. The CapEx guide is also a range that tries to reflect that uncertainty, both in terms of what we find we may need, including for our Gen AI launches and the features that we talked about at Connect, which are all very new and many of which are only available in limited markets today. So we’re going to learn -- 7 of 15 -- 8 a lot more about what our inference needs might look like over time, and then as well as how market conditions might evolve. Chad Heaton: Yes, so on your second question about just giving more color on the hiring and backlog commentary, I think there’s a couple points we’re trying to land here. So, I think the first is the ethos of the year of efficiency is continuing. And so, we’re in the middle of our budget process right now, we mentioned that we’re looking to invest in AI in particular, but we’re also looking to offset a lot of that investment through reductions in the non-AI portions of the business. And so, that’s sort of a key message from a budget perspective, but we also want to note that we are at a hiring underrun currently relative to our 2023 budgets, and so we do expect some catch-up there to happen over the course of 2024. So on the one hand, we expect that to ramp, but we’re also saying from a budgeting philosophy perspective, we’re looking to sort of offset the investments with this sort of continued focus on efficiency going forward. Ken Gawrelski: Thank you very much. Operator: Your next question comes from the line of Rob Sanderson with Loop Capital. Rob Sanderson: Hey, thank you very much. Two questions from me as well. Susan, you mentioned that Reels will become a more modest tailwind in ’24, and presumably a fairly material tailwind through ’23, as you’ve got to work your way all the way back to revenue neutrality. Much of the focus of Reels has been geared towards Instagram, so you -- do you see the comparable opportunity for Reels on Facebook and how far into tapping that opportunity do you think you are compared to Instagram? That’s question one, and question two, on messaging products, is there anything notable -- any notable differences in cost structure, particularly cost of revenue versus your sort of more workhorse ad units and Feed and Stories and the like? And this was sort of asked on the main call, but we’re just -- I’m trying to think through the puts and takes from introducing more Gen AI into messaging and certainly heavier infrastructure cost component is part of it, -- 8 of 15 -- 9 but could be a material catalyst for adoption by businesses, and driving auction dynamics in ARPU and all that flywheel. So, I recognize it’s real early and there’s lots of unknowns. I’m kind of fishing for the things that we might know or that we can reasonably assume, and things to think about as this evolves from a probability perspective. Thank you. Susan Li: Thanks, Rob. I think -- so on your first question about Reels, it was a -- and to be clear, it was a headwind on the net, for 2023, was a headwind for most of the year, so it’s just reached revenue neutrality. And as you mentioned, Instagram is a little bit ahead of Facebook on -- well, on Reels, period. But it is -- I think in both cases, again, we’ve been very pleased with both the steady progress that we’ve made making the consumer experience more engaging and more incremental in terms of its contribution to overall engagement, and then of course improving monetization efficiency. And we expect that this is -- that that work will continue into ’24, where because it will be a tailwind and not a headwind anymore, in terms of net revenue impact for the year, that will certainly be incremental to the 2024 revenue outlook, but we also really -- I think -- the way I would think about this, and Mark, I believe, used the language maybe that it graduated. Reels is self sufficient now. It is now really a part of the core experience of both Instagram and Facebook, and it is part of now in our sort of broader video ecosystem. It is one of a number of video formats that we recognized playing an increasingly important role on the platform, especially on Facebook where there is a sort of more diverse set of video formats including longer-form video, including live streaming, et cetera. So, we’ll be continuing to invest in Reels as well as the video ecosystem more broadly, and we’ll be managing sort of the engagement and monetization goals holistically in balance with each other. We expect it will contribute to revenue in 2024, and we don't expect to quantify that. -- 9 of 15 -- 10 Your second question, which is around -- I’m going to do my best to answer your second question, but you should tell me if I’m off base here. I think what you're alluding to is over the long-term do we think the cost structure of -- you said -- your question was about messaging, but I assume you’re really talking about like Gen AI-powered messaging because you were comparing it to Feed. I think we just -- we don't know yet what the longer-term sort of cost profile is going to look like here in part because we are a lot earlier in the cost efficiency curve for our GPU utilization than we are for CPUs. And so, I think on the CPU side, we have certainly brought -- we’ve certainly achieved a lot of cost efficiencies overtime, in 2023 in particular. We’ve done that through a combination of deployment of our in-house MTIA customs systems and improved performance and cost, especially for the recommendation work, which is a lot of what powers some of the core experiences you were talking about, and then optimizing our models to deliver better results with fewer resources. And we’ve also just benefited from improvements in commercially-available technology. On the Gen AI side, we are just earlier. Over the long-term, we expect Gen AI to follow a similar trajectory as we saw in our ranking and recommendations work and to have the opportunity to deliver more cost efficiencies there. But you're right that it is more compute heavy today, and we’re just earlier in the development of the technology and really more focused right now on developing and supporting some of the new compelling consumer and business-facing capabilities that we’ve assessed. Operator: And your next question comes from the line of Rohit Kulkarni with MKM Partners. Rohit Kulkarni: Hey. Thanks. Two questions. One’s on Reels, and just any unit of measures of profitability for Reels versus the business? Do you feel there are any structural differences between kind of revenue generated from Reels per unit of time or any unit of engagement? It’s very different from other parts -- 10 of 15 -- 11 of Newsfeed that you think Reels can be pushed even more from a unit economics perspective? And then -- and second question, whether this kind of narrative that the number of days between Thanksgiving and Christmas is the highest this year and perhaps some of the ad spend is going to be delayed by some of the customers, just wondering if you are seeing any such dynamic between how some advertisers may be delaying ad spend? And perhaps that’s leading to some of the slow start to 4Q? Susan Li: Thanks, Rohit. So, on your first question, we have not broken out Reels- specific profitability measures. Having said that, we have talked about one of the more fundamental structural supply constraints in terms of the opportunities to show ads on Reels is simply the difference in scroll speed in Reels and -- versus non-video formats. And so, we do expect that that structural difference will persist. And it’s a place where we have and are continuing to invest in work to find more creative ways to show Reels on ads in order to mitigate that particular issue. But we have talked about Reels being sort of lower monetization per time for this reason, and that’s something that we are investing in sort of all facets of finding ways to increase supply and then, of course, finding ways to make Reels ads more performant and more interactive over time. So, that’s the -- that’s the Reels question. On your second question about the -- are advertisers delaying ads as a result, which is playing into a slower start to Q4? Seasonality has varied by year. It’s obviously very difficult to know perfectly in advance. But I don’t think that we specifically expect any particular difference in spend patterns this year due to the timing issues that you described. Operator: And your next question comes from the line of Brent Thill with Jefferies. Brent Thill: Thanks. On the Advantage+, can you just speak to the roll-out broadly across the platform? Ultimately, what you’re seeing so far? And I had a quick follow up. -- 11 of 15 -- 12 Susan Li: Yes. Well, Advantage+ is something we love talking about. We talk about how it has scaled to a $10 billion annual run-rate since launching it last year. I alluded to this on the call, which is just that we’ve been continuing to roll out new features for the Advantage+ Suite of tools. In Q3, we introduced features based on advertiser feedback and input about things that they want to have control over. So, we’ve launched things like ad scheduling, we’ve introduced the ability to for advertisers to set and optimize for a minimum ROAS. We continue to have other feature launches in the Advantage suite of tools. I think we’ve talked about some of them, the Advantage+ audience tools that help sort of build the right audience creation for advertisers, Advantage+ catalog ads, which are especially useful sort of in the e-commerce vertical, Advantage+ creative, which is the set of creative enhancements to optimize images in creative -- images and text in their ad creative, and then increasingly bringing the generative AI tools that we tested out in the AI sandbox into the Advantage+ creative features. So, we’ve really introduced a lot of new ad products and features over the last year. We’re glad a lot of them are in market, in particular, in time for the holiday season. And, of course, the fact that Advantage+ shopping really began as an ads product sort of initially targeted at e-commerce and retail advertisers, I think, means that we’re particularly well suited to benefit from the advertisers using it to drive sales over the holiday marketing period. Brett Thill: And just a quick follow up on the CapEx. I think everyone was braced for a slightly larger number. Your peers have been putting up some pretty big numbers and I think maybe the explanation is you’ve already invested in AI and you have a lot of capabilities in, and maybe incrementally you don’t need to add as much. -- 12 of 15 -- 13 I mean, I know there’s capacity constraints you spoke about as well but any other kind of color to this counter to what hearing from what we’re hearing from some of your other tech peers? Susan Li: We really -- I mean we are growing overall CapEx investments next year. We’re growing our overall AI investments next year. We’re investing more in AI hardware and in building data centers with the new data center architecture we talked about, which will be AI enabled but have the flexibility to be used for non AI workloads as well. We have already stood up significant training capacity for generative AI this year. Next year, a lot of the investment will be across both training and inference needs and obviously on the inference side it’s a bit harder to predict. So, we believe we can meet our capacity needs with our planned investments, but obviously we will learn more about that overtime. We also have flexibility since we’re using systems that can support both training and inference. And it is just a very dynamic area of planning. There are a lot of different puts and takes on both the demand planning side, including the things we know, as well as the things we expect to learn over the next year and then, of course, the supply constraints and market conditions. So it’s an area that we expect will evolve next year. That’s why we have a pretty wide range to start with and we expect that we will learn more through the course of 2024 in terms of where we actually expect that we’ll be -- expect that we’ll need and what we’ll be able to land. Kenneth Dorell: Dave, we have time for one last question. Operator: Certainly, thank you. That will come from the line of Michael Nathanson with Moffett Nathanson. Your line is open. Michael Nathanson: Hey, thanks. Susan, just following up on Brett’s two questions. I had -- one’s teed up. One is Advantage+, is there any way to talk about the incrementality of the $10 billion revenue run rate? How much of that you think is incremental? And then Advantage+, on ROAS. How would it compare -- 13 of 15 -- 14 to previous products, pre-IDFA in terms of ROAS, are you guys back to where you were pre-IDFA changes? And then on CapEx, historically you’ve been lumpy on CapEx where you’ve built data centers ahead of demand. But I wonder on this new architecture, how does that change maybe the capital intensity over business cycle versus prior build outs? Thanks. Susan Li: On your first question, which was about Advantage+, we haven’t talked about the incrementality specifically. We are at a place now where we are releasing a lot of new features into the Advantage+ suite of tools. So, we expect that those -- a lot of those features and tools are driving the ongoing performance gains that we’ve seen for advertisers and that we have -- we’ve been able to increase the conversions that we’re delivering and we think that advertisers are seeing increased spend on the platform. So, we’re not quantifying the incrementality but I think generally we certainly believe that the Advantage+ set of tools is delivering incremental value. The second part of that first question, I think, was kind of where are you on ROAS as relative to the pre-ATT world. And I don’t know. We don’t know what the landscape would like today without the ATT changes. We certainly believe that advertizing is a relevant performance game and that we have made and continue to make improvements to our performance -- to the features available, to the performance and measurement of our ad products in recent periods and we’ve received consistent positive feedback from advertisers on that. So, we feel good about our position relative to other options for advertisers in the market right now. The second question was on CapEx. You said, historically, we are -- we’ve been lumpy on CapEx as we build DC, data center architecture, ahead of demand. How does this change? Michael Nathanson: Right. -- 14 of 15 -- 15 Susan Li: Certainly, our year-over-year CapEx growth comes from -- I mean, it really comes from multiple drivers. So, obviously, we expect that there is going to be meaningful server growth, too, right? And the server growth is across both non-AI and AI workloads. And then, we are expecting to ramp up construction on the new data center architecture in 2024 and to sort of ramp up more sites. Will it be -- I’m not sure if it will be lumpier. I think this is something we’ll have to evaluate to sort of get data center versus server needs obviously over the course of the coming years. We do try to build data center capacity ahead of server needs. But we also think that the higher flexibility of the next-generation data center architecture should, again, enable us more flexibility in the planning going forward. It would have been lumpier without, effectively, if we needed to build ahead of both AI and non-AI demands. Kenneth Dorell: Great. Thank you for joining today. We appreciate your time, and we look forward to speaking with you all again soon. Operator: And this concludes today's conference call. Thank you for joining us. You may now disconnect your lines. -- 15 of 15 --