Good afternoon. My name is Krista, and I will be your conference operator today. At this time, I would like to welcome everyone to the Meta First Quarter Earnings Conference Call. This call will be recorded. Thank you very much.

Thank you. Good afternoon, and welcome to Meta Platform's First Quarter 2021 Earnings Conference Call. Joining me today to discuss our results are Mark Zuckerberg, CEO; and Susan Li, CFO. 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 earnings press release and in our annual report on Form 10-K 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 earnings 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 over to Mark.

All right. Thanks, Ken, and everyone, thanks for joining. It's been a good start to the year, both in terms of product momentum and business performance. We estimate that more than 3.2 billion people use at least one of our apps each day, and we're seeing healthy growth in the U.S. I want to call out WhatsApp specifically, where the number of daily active users and messages sent in the U.S. keeps gaining momentum, and I think we're on a good path there. We've also made good progress on our AI and metaverse efforts, which will be the focus of my comments today. So let's start with AI. We're building a number of different AI services, from Meta AI, our AI assistant that you can ask any question across our apps and glasses, to creator AIs that help creators engage their communities that fans can interact with, to business AIs that we think every business on our platform will eventually use to help customers buy things and get customer support. We're also developing internal coding and development AIs, as well as hardware like glasses for people to interact with AIs, and a lot more. Last week, we had the major release of our new version of Meta AI, which is now powered by our latest model, Llama 3. Our goal with Meta AI is to build the world's leading AI service in terms of both quality and usage. The initial rollout of Meta AI is going well. Tens of millions of people have tried it and feedback is very positive. When I checked in with our teams, the majority of feedback was that people wanted us to release Meta AI for them wherever they are. So we've started launching Meta AI in some English-speaking countries and will roll out in more languages and countries over the coming months. You all know our product development playbook by this point. We release an early version of a product to a limited audience to gather feedback and start improving it, and then once we think it's ready, we make it available to more people. That early release was last fall, and with this release, we are now moving to the next growth phase of our playbook. We believe Meta AI with Llama 3 is now the most intelligent AI assistant that you can freely use. We're making it easier for a lot of people to use it within WhatsApp, Messenger, Instagram, and Facebook. In addition to answering more complex queries, a few other unique features from this release include Meta AI's ability to create animations from still images and generate high-quality images quickly—so fast that it can create and update them as you type. I've seen many people commenting about that experience online, expressing how they've never experienced anything like it before. Regarding the core AI model and intelligence powering Meta AI, I'm very pleased with how Llama 3 has come together so far. The 8 billion and 70 billion parameter models we released are best-in-class for their scale. The 400-plus billion parameter model we’re still training appears on track to be industry-leading on several benchmarks. I expect that our models will only improve further with open-source contributions. Overall, I view the results our teams have achieved here as a key milestone in demonstrating that we have the talent, data, and ability to scale infrastructure to build the world's leading AI models and services. This leads me to believe that we should invest significantly more over the coming years to build even more advanced models and the largest AI services in the world. As we scale CapEx and energy expenses for AI, we'll continue focusing on operating the rest of our company efficiently. Realistically, even with directing much of our existing resources to focus on AI, we'll still grow our investment envelope meaningfully before we realize revenue from these new products. It's worth mentioning that we've historically seen volatility in our stock during this phase of our product playbook, when we invest in scaling a new product but aren't yet monetizing it. We've seen this with Reels and Stories as newsfeed transitioned to mobile, among others. I also expect a multi-year investment cycle before we fully scale Meta AI, business AIs, and more into the profitable services I anticipate. Historically, investing to build these new scaled experiences in our apps has been a very good long-term investment for us and for investors who have stayed with us. The initial signs are quite positive here, too. However, building the leading AI will be a larger undertaking than previous experiences we've added to our apps, likely requiring several years. Once our new AI services reach scale, we have a strong track record of monetizing them effectively. There are several ways to build a massive business here, including scaling business messaging, introducing ads or paid content into AI interactions, and enabling people to pay for larger AI models and more compute access. Additionally, AI is already helping us improve app engagement, leading to more ads and improved ad delivery to offer more value. If technology and products evolve as we hope, each of these factors will unlock massive amounts of value for people and businesses over time. We're seeing solid progress on these efforts already. Currently, about 30% of posts in Facebook feeds are delivered by our AI recommendation system, up 2x from the previous couple of years. For the first time, more than 50% of the content users see on Instagram is AI-recommended. AI has also been integral in enhancing the value we create for advertisers by showing users more relevant ads. If you consider our two end-to-end AI-powered tools, Advantage+ shopping and Advantage+ app campaigns, revenue from those tools has more than doubled compared to last year. We'll continue to remain focused on efficiency as we scale Meta AI and other AI services. Some efficiencies will come from improving how we train and run models, while some improvements will stem from the open-source community, particularly in cost efficiency as we saw with Open Compute. We're also making progress on developing more of our silicon. Our Meta training and inference accelerator chip has successfully enabled us to run some of our recommendations-related workloads on this cheaper stack. As this program matures over the upcoming years, we aim to expand this to more of our workloads as well. Of course, as we ramp these investments, we will also continue to manage headcount and other expense growth throughout the company carefully. Beyond our AI work, another focus is the metaverse. It's intriguing how these two themes have converged. This is clearest when we look at glasses. I used to think AR glasses wouldn't become a mainstream product until we developed full holographic displays—and I still think that's going to be awesome. However, it now seems evident that there's also a meaningful market for fashionable AI glasses without a display. Glasses provide an ideal device for an AI assistant because they can perceive what you see and hear, offering full context on your surroundings as they assist with your tasks. The launch this week of Meta AI integrating Vision on the glasses is a good example, allowing you to ask questions about what you’re looking at. One strategic dynamic I've been considering is that an increasing amount of our Reality Labs work is shifting toward supporting our AI efforts. We currently report our financials as two distinct segments—Family of Apps and Reality Labs—but strategically, I consider them fundamentally the same business. Reality Labs' vision is to build the next generation of computing platforms, primarily to establish the best apps and experiences on top of them. Over time, we will need to find better ways to articulate the value generated across both segments, so it doesn't just appear that our hardware costs are increasing as our glasses ecosystem scales with all value flowing to a different segment. The Ray-Ban Meta glasses built in partnership with Essilor Luxottica continue to perform well and are sold out in various styles and colors. We're working to make more and release additional styles as quickly as possible. We just launched the new cat-eye Skyler design yesterday, which is more feminine, and generally, I'm optimistic about our plan of starting with classic designs and expanding with a growing diversity of options over time. If we want everyone to access wearable AI, eyewear differs from phones or watches because people desire diverse designs. Our collaboration with leading eyewear brands will help us address a larger market. I believe a similar open ecosystem approach will aid us in expanding the virtual and mixed reality headset market over time as well. We announced the opening of Meta Horizon OS, the operating system powering Quest. As the ecosystem grows, I believe there will be a sufficient diversity in how individuals utilize mixed reality, driving demand for more designs than we can build in-house. For instance, a work-focused headset may require specific design attributes to connect to a laptop, while a fitness-focused headset might prioritize lightweight, sweat-wicking materials. An entertainment-focused headset might emphasize the highest resolution displays, while a gaming-focused headset could focus on peripherals and haptics, possibly bundled with Xbox controllers and a game pass subscription. I think that our first-party Quest devices will remain the most popular headsets, and we will continue to focus on advancing state-of-the-art technology and making it accessible to all. However, I also believe that opening our ecosystem and operating system will accelerate the growth of the broader mixed reality ecosystem. We're also observing positive improvements across our apps. I have mentioned some of the most important trends already, including WhatsApp growth in the U.S. and AI-powered recommendations in our feeds and reels. However, I want to highlight that video remains a bright spot. This month, we launched an updated full-screen video player on Facebook, unifying reels, longer videos, and live content into a single experience with a cohesive recommendation system. On Instagram, reels and video continue to drive engagement, with reels now constituting 50% of the time spent within the app. Threads is also seeing solid growth, with over 150 million monthly active users; we're on track with the trajectory I hoped for. And, of course, my daughters would want me to mention that Taylor Swift is now on Threads, a significant development in my household. Overall, I am proud of the progress we've made thus far this year. We have much more execution ahead to capitalize on the opportunities before us. A huge thank you to our teams for driving these advancements and to all of you for being on this journey with us. Now here is Susan.

Thanks, Mark, and good afternoon, everyone. Let's begin with our consolidated results. All comparisons are on a year-over-year basis unless otherwise noted. Q1 total revenue was $36.5 billion, reflecting a 27% increase on both a reported and constant currency basis. Q1 total expenses were $22.6 billion, up 6% compared to last year. In terms of specific line items, cost of revenue increased 9% due mainly to higher infrastructure-related costs, partially offset by lapping Reality Labs' inventory-related valuation adjustments. R&D grew 6%, primarily driven by higher headcount-related expenses and infrastructure costs, offset by lower restructuring costs. Marketing and sales expenses decreased 16% mainly due to reduced restructuring costs, professional services, and marketing expenditure. General and administrative expenses increased by 20%, as higher legal-related expenses were partially offset by lower restructuring costs. We closed the first quarter with over 69,300 employees, up 3% from Q4. First quarter operating income reached $13.8 billion, representing a 38% operating margin. Our tax rate for the quarter was 13%, with a net income of $12.4 billion, or $4.71 per share. Capital expenditures, including principal payments on finance leases, totaled $6.7 billion, driven by investments in servers, data centers, and network infrastructure. Free cash flow stood at $12.5 billion, and we repurchased $14.6 billion of our Class A common stock, along with $1.3 billion in dividends to shareholders, concluding the quarter with $58.1 billion in cash and marketable securities and $18.4 billion in debt. Moving to our segment results, I'll start with our Family of Apps segment. Our community across the Family of Apps continues to grow, with about 3.2 billion people using at least one of our apps daily in March. Q1 total Family of Apps revenue was $36 billion, a 27% year-over-year increase. Family of Apps ad revenue was $35.6 billion, also up 27% or 26% on a constant currency basis. Within ad revenue, the online commerce vertical was the largest contributor to year-over-year growth, followed by gaming and entertainment and media. Geographically, ad revenue growth was strongest in the Rest of World and Europe at 40% and 33%, respectively. Asia Pacific grew 25%, and North America grew 22%. In Q1, we saw a 20% increase in total ad impressions served across our services, while the average price per ad rose by 6%. Impression growth was mainly fueled by Asia Pacific and Rest of World, while pricing growth was driven by advertiser demand, which was partially offset by strong impression growth, particularly from lower-monetizing regions and services. Family of Apps other revenue was $380 million in Q1, up 85%, driven by business messaging revenue growth on our WhatsApp business platform. We're continuing to direct the majority of our investments toward the development and operation of our Family of Apps. In Q1, Family of Apps expenses totaled $18.4 billion, accounting for approximately 81% of our overall expenses, which were up 7% due mainly to higher legal and infrastructure costs, partially offset by lower restructuring costs. Family of Apps operating income was $17.7 billion, reflecting a 49% operating margin. Within our Reality Labs segment, Q1 revenue was $440 million, up 30%, driven mainly by Quest headset sales. Reality Labs expenses were $4.3 billion, down 1% year-over-year as higher headcount-related expenses were significantly offset by inventory-related valuation adjustments and restructuring costs. Reality Labs operating loss was $3.8 billion. Turning to our business outlook, two primary factors drive our revenue performance: our ability to deliver engaging experiences for our community and our effectiveness at monetizing that engagement over time. We're pleased with our engagement trends and have strong momentum across our product priorities. Our investments in developing increasingly advanced recommendation systems continue to enhance engagement on our platform, demonstrating that people find added value in discovering content from accounts they're not connected to. The level of recommended content in our apps has scaled as we've improved these systems, and we see further opportunities to increase relevance and personalization of recommendations as we enhance our models. Video continues to grow across our platform, now representing more than 60% of total time on both Facebook and Instagram. Reels remains the primary driver of that growth, and we are progressing on our initiative to unify Reels, longer-form video, and live video into a single experience on Facebook. In April, we rolled out this unified video experience in the U.S. and Canada, increasingly powered by our next-generation ranking architecture, which we expect will yield more relevant video recommendations over time. We're also introducing deeper integrations of generative AI into our apps in the U.S. and over a dozen other countries. Along with using Meta AI within our chat surfaces, people will now be able to use Meta AI in search within our apps, as well as in Facebook feeds and groups. We expect these integrations will support our social discovery strategy as our recommendation systems help people explore their interests while Meta AI allows them to delve deeper into topics of interest. Threads continues to gain traction, as we keep shipping valuable features and scaling the community. Now for the second driver of our revenue performance: increasing monetization efficiency. There are two components to this work. The first focuses on optimizing ad levels within organic engagement. Here, we continue to enhance our understanding of user preferences for viewing ads, allowing us to optimize the right time, place, and audience to show an ad. For instance, we are improving our ability to adjust ad placement and quantity in real-time based on user interest and ad content to minimize disruption from ads while innovating on new and creative ad formats. We expect to continue this work while prioritizing surfaces with relatively lower monetization levels, such as video and messaging, as additional growth opportunities. The second component involves enhancing marketing performance. AI is playing an increasingly central role in these efforts. We are continuously improving our ad modeling to deliver superior performance for advertisers. One example is our new ads ranking architecture, Meta Lattice, which we rolled out broadly last year. This architecture allows us to operate significantly larger models that generalize learnings across objectives and surfaces, replacing numerous smaller, specialized ad models that were previously optimized for individual objectives. This leads to enhanced efficiency by operating fewer models while simultaneously improving ad performance. Another way we leverage AI is by providing greater automation for advertisers. Through our Advantage+ portfolio, advertisers can automate aspects of the campaign setup process, such as selecting ad creatives, or fully automate their campaigns using our end-to-end automation tools, Advantage+ shopping and Advantage+ app ads. We are witnessing increasing adoption of these solutions and anticipate driving further uptake throughout 2024 while applying insights gained to our broader ads investments. Now, turning to our approach to capital allocation. We continue to see compelling investment opportunities that will not only enhance our core business in the near term but also help us capture substantial long-term opportunities in generative AI and Reality Labs. As we develop more advanced and compute-intensive recommendation models and scale our capacity for generative AI training and inference needs, we believe ensuring adequate infrastructure capacity will be crucial in realizing many of these opportunities. Therefore, we intend to significantly increase infrastructure investments over the coming years. We are also making substantial ongoing investments in Reality Labs. Our AI initiatives have increasingly intersected with our Reality Labs efforts. For instance, with Ray-Ban Meta smart glasses, individuals in the U.S. and Canada can now use our multimodal Meta AI assistant for daily tasks without needing to pull out their phones. In the long run, we expect generative AI to play a more significant role in our mixed reality products, simplifying the development of immersive experiences. Expanding our AI initiatives will help ensure we provide the optimal version of our services as we transition to the next computing platform. We plan to pursue these opportunities while maintaining a focus on operating discipline and believe our robust financial position will enable us to support these investments while also returning capital to shareholders through share repurchases and dividends. Moreover, we continue to monitor a dynamic regulatory landscape, including potential regulatory headwinds in the EU and U.S. that may significantly affect our business and financial results. We also have a jury trial scheduled for June concerning a lawsuit initiated by the state of Texas regarding our use of facial recognition technology, which could lead to a material loss. Now, to the revenue outlook: we expect second quarter 2024 total revenue to range from $36.5 billion to $39 billion. Our guidance assumes foreign currency presents a 1% headwind to year-over-year total revenue growth based on current exchange rates. Turning to the expense outlook: we expect full year 2024 total expenses to range from $96 million to $99 billion, updated from our previous outlook of $94 million to $99 billion due to increased infrastructure and legal costs. For Reality Labs, we anticipate operating losses will rise significantly year-over-year due to our ongoing product development efforts and investments to scale our ecosystem. For the CapEx outlook, we foresee our full year 2024 capital expenditures to be between $35 billion and $40 billion, increased from our previous range of $30 billion to $37 billion as we accelerate investments in infrastructure to support our AI roadmap. Although we aren't issuing guidance beyond 2024, we expect CapEx to continue to rise next year as we invest heavily to support our ambitious AI research and product development initiatives. Regarding taxes, barring changes to our tax landscape, we anticipate our full year 2024 tax rate will fall in the mid-teens. In conclusion, Q1 marked a positive start to the year. We are experiencing strong momentum within our Family of Apps and making significant progress on our long-term AI and Reality Labs initiatives, which have the potential to transform how people interact with our services in the coming years.

And your first question comes from the line of Eric Sheridan from Goldman Sachs.

Maybe I'll ask a two-parter. Mark, you used the analogy of other investment cycles you've been through around products like Stories and Reels. I know you're not giving long-term guidance today, but using those analogies, how should investors think about the length and depth of this investment cycle regarding AI and/or Reality Labs more broadly and mixed reality? And you both referenced AI's impact on the advertising ecosystem. What are you monitoring in terms of adoption or utility on the consumer side to gauge whether AI adoption is tracking alongside the investment cycle?

Yes. Regarding the timing, it's somewhat challenging to extrapolate from previous cycles. The key takeaway is that it usually takes us a few years, potentially a little more or a little less, to focus on building out and scaling products. Typically, we don't prioritize monetization of new areas until they achieve significant scale, as it's much more effective to improve monetization on other aspects before these new products reach scale. We enter a phase where savvy investors see that the product is scaling and that there is a clear monetizable opportunity even before revenue materializes. We've seen this with Reels and Stories and the transition to mobile, where we first build out the inventory and then monetize it. Sometimes, during the scaling phase of a product, it's not only the case that we're not generating income from it, but it can also displace revenue generated from other areas. This was evident with Reels, where it scaled, but there was a period when it wasn't profitable until it became so later. This analogy is a reminder that we should look at the consumer products' scaling. Meta AI has recently launched meaningfully, so we don't yet have solid statistics to share on its performance. For this year, and probably much of next year, my focus remains on growing Meta AI and other AI products and their engagement. If those products are on a solid growth trajectory, they will likely evolve into very substantial businesses.

Your next question comes from the line of Brian Nowak from Morgan Stanley.

Thanks for taking my questions. I have two. The first one is regarding the recommendation engine improvements. Susan, you mentioned further opportunities to increase the relevance of the models. Could you unpack that a bit? Can you provide examples of where you're still running the model on a suboptimal basis or opportunities for improved signal capture or data utilization? What areas do you see for improvement from here? And then on the second topic, regarding driving incremental adoption of AI tools for advertisers, what are the main barriers you've encountered to encourage advertisers to test these tools? How do you plan to address this throughout 2024 and 2025?

Thanks, Brian. To your first question about leveraging and improving our recommendation models to boost engagement, historically, each recommendation product, including Reels and in-feed recommendations, has utilized its own AI model. Recently, we've started developing a new model architecture designed to power multiple recommendation products. We began partially validating this model last year using it for Facebook Reels, where we saw meaningful performance gains, with an 8% to 10% increase in watch time. This year, we plan to extend the singular model architecture to recommend content across Facebook Reels, and its video tab as well. While it’s still early to share specific results, we are optimistic that this new model architecture will lead to increasingly relevant video recommendations in time. On the advertising side, we discussed the new model architecture, Meta Lattice, that we rolled out last year, which consolidates smaller, specialized models into larger ones that can learn what characteristics improve ad performance across multiple services. This has led to enhanced ad performance as we deployed it across Facebook and Instagram in support of various objectives. In 2024, we aim to further enhance model performance and support even more objectives like web, app and ROAS. We're investing significantly in the underlying model architecture for organic engagement and ads, expecting to see continuous improvements in ad performance. Regarding your second question about getting advertisers to try and adopt generative AI tools, we have two aspects. The short-term focus is on generative AI ad creative features we’ve integrated into our ad creation tools. The adoption of these features has varied across verticals and advertiser sizes. Notably, we’ve seen considerable uptake of image expansion among small businesses, and this will remain a significant priority for us in 2024. I expect enhancements to our underlying foundational models will improve the quality of generated outputs and support new features. Currently, we have features supporting text variations, image expansion, and background generation, and we're working continuously to enhance their performance for advertisers to create personalized ads at scale. The long-term focus is on business AIs. We are testing the ability for businesses to implement AIs for business messaging that represent them in customer chats—starting with shopping use cases, such as responding to inquiries about a product or its availability. We are in early stages, testing this with a handful of businesses on Messenger and WhatsApp, receiving positive feedback from businesses claiming the AIs saved them considerable time, while consumers noted quicker response times. We're learning from these tests to enhance AI performance over time and will expand these tests over the coming months, ensuring we address the requirements effectively before broader availability.

Your next question comes from the line of Mark Shmulik from Bernstein Research.

I guess back to that product playbook we've discussed a few times. With Reels now commanding a substantial share of time spent on Instagram and Facebook, how should we regard the next phase of monetization growth from here? Specifically, as we pivot back to platform shopping or other monetization strategies, can you share any insights regarding our roadmap beyond ad insertion? And Susan, on the ad market, specifically, we previously heard a lot about the contribution from Chinese-based advertisers. Can you provide any updates on how that spending is trending?

Sure. Thanks, Mark. Reels revenue continues to grow across Instagram and Facebook in Q1, driven by heightened engagement and enhanced monetization efficiency through improvements in our ads ranking and delivery. As stated previously, we don't plan on quantifying the impact from Reels going forward but it remains a valuable contributor to our overall revenue. We expect ongoing performance improvements and growth in supply. On performance improvements, we are investing in continuous ranking enhancements and making ads easier to interact with through efforts like optimizing calls to action and post-click experiences, especially crucial for direct response performance. We’re also optimizing ads to feel more native to Reels. In Q1, we rolled out our generative AI image expansion tools across Facebook and Instagram Reels after its introduction in Instagram feed in Q4, seeing considerable adoption from small businesses. We’re optimistic about the potential to keep enhancing ad performance. Despite an increase in Reels ad loads over the past year, it remains lower on a per-time basis than both Feed and Stories. We will continue to seek opportunities to grow it thoughtfully and invest in inventive ways to address structural supply constraints of Reels, especially considering its video-heavy format, including implementing higher density experiences and formats and increasing ad load personalization to ensure we're reaching people when they're most likely to be engaged. Regarding China, growth in spending from Chinese advertisers remained strong in Q1, predominantly fueled by online commerce and gaming sectors. This trend is evident in the Asia Pacific advertising segment, which was the fastest-growing region at 41% year-over-year in Q1. We witnessed robust growth across other geographies as well, including a 6-point acceleration in total revenue growth from North American advertisers. While we aren’t quantifying Q1 contributions from China, we recognize we’re lapping strong demand periods anticipated throughout 2024, given the recovery of China-based advertisers from prior pandemic-driven obstacles.

Your next question comes from the line of Doug Anmuth from JPMorgan.

Can you discuss what changes most in your perspective regarding the business and the opportunities now compared to 3 months ago? Are there any particular revenue cautions in the ad market? Is the AI opportunity even larger, thus necessitating more investment than originally anticipated?

Yes, I'm increasingly optimistic and ambitious about AI. Previously, while looking at last year, when we released Llama 2, we were excited about the model and believed it would serve as a basis for building several valuable integrations into our social products. Now, we are in a fundamentally different position. With our latest models, we’re not only building capable AI models for new social and commerce applications but are demonstrating our ability to develop leading models and position ourselves as a prominent AI company. This revelation unveils additional opportunities beyond the most visible ones for us. This was the point in my initial remarks where I mentioned the technical validation stemming from the success of Llama 3 and Meta AI illustrates that we have the talent, data, and infrastructure to scale leading work in AI. With our focus on Meta AI, I envision it becoming the most utilized and top-performing AI assistant globally—a development that will prove immensely valuable. This progress encourages me to invest further to remain at the forefront of this industry. We're currently scaling the product even before it begins generating revenue, akin to the analogy I made earlier regarding historical investment cycles we've undergone. However, if you observe the output of our team, it's apparent that our optimism and ambition have increased significantly, suggesting these products will be vital for us.

In response to the second question, we're not issuing full-year 2024 guidance. Moreover, our revenue for the entire year will be influenced by numerous factors, including macro conditions that become harder to predict as the forecast extends further out. Throughout 2024, we will also be facing tougher comparisons due to significant demand levels. That said, we believe we can continue cultivating engagement and growth across our products through investments in AI-based content recommendations and ongoing video initiatives. We are also expecting to enhance our ads' performance and deliver greater value to advertisers. A notable trend is that we saw our conversion rates increase at a faster rate than the growth in our impressions this past quarter. This indicates that our conversion grade is improving—one of the many ways in which our ads are becoming more effective.

Your next question comes from the line of Justin Post from Bank of America.

First on the CapEx front, you’re discussing an investment cycle. Can you shift some of the Metaverse spending toward AI? Are there areas where they converge that would allow you to fund AI through the Metaverse budget? And second, long-term investors are increasingly focused on returns on capital. Given the strong returns from CapEx in the past, how should we think about the returns on the capital expenditures you're making? What is your outlook for these returns two or three years out?

To address the second part, regarding measuring the ROI on our CapEx investments, we've broadly categorized our AI investments into two categories: core AI work and strategic bets, which encompasses generative AI and advanced research efforts. These categories are at different stages concerning ROI measurement and driving revenue for our business. For our core AI work, we maintain a strongly ROI-driven investment approach, observing robust returns as enhancements in engagement and ad performance translate into revenue growth. In contrast, the strategic bets category is still in the early stages. Mark has referenced the potential to create meaningful value for our business across multiple areas, necessitating upfront investments to develop advanced models and increase product usage before they yield significant revenue. While the long-term potential is enormous, progress in creating measurable returns is significantly earlier than with core AI work. Importantly, we’re building our systems to provide us flexibility for resource allocation, allowing the capacity to be utilized across various applications as ideal opportunities emerge.

Addressing the question about reallocating resources from other company segments, we’re broadly adjusting resources across various areas, whether they involve compute resources or others, to advance our AI efforts. For Reality Labs, I remain optimistic about developing these new computing platforms for the long term. I mentioned earlier that a primary focus of our investment in Reality Labs is glasses. We believe this will be a significant platform for the future. Our outlook has improved substantially, as we previously thought we would need to wait for full holographic display technology to enter the market significantly. We are now much more focused on the glasses we are developing in partnership with Ray-Ban, which are performing well. Consequently, Reality Labs is increasingly aligned with our AI goals. Still, I hold firm that we must also concentrate on building these long-term platforms.

Your next question comes from the line of Youssef Squali from Truist Securities.

Mark, in light of the recent legislation regarding the potential ban or sale of TikTok, how do you anticipate this will affect the U.S. social media landscape? Additionally, what would you say to those who believe this could lead to the government selecting winners and losers? Susan, can you share how significant Advantage+ is in terms of spending on the platform and its impact on stabilizing overall CPMs?

We've been closely watching developments surrounding TikTok, but at this point, it's premature to evaluate its possible impacts or implications for our business. Regarding Advantage+, we continue seeing solid traction across our Advantage+ portfolio, including both solutions that automate individual steps of campaign creation and those that provide full end-to-end automation. For example, the single-step automation product, Advantage+ audience, has experienced significant growth in adoption since we made it the default setting for most advertisers in Q4. This allows advertisers to enhance campaign performance by utilizing audience inputs as suggestions rather than strict parameters. Based on testing, campaigns using Advantage+ audience targeting have seen an average 28% reduction in cost per click compared to traditional targeting methods. For comprehensive automation products such as Advantage+ shopping and Advantage+ app campaigns, we have also witnessed robust growth. Mark noted that the combined revenue from those two has more than doubled since last year. We believe significant opportunities remain to broaden adoption, as we aim to facilitate more conversion types for Advantage+ shopping, recently adding ten more types beyond purchase events. We are consistently integrating more functionalities into the Advantage+ tools over time, especially where many of our generative AI ad creative features have been incorporated, providing advertisers opportunities to experiment.

Your next question comes from the line of Ken Gawrelski from Wells Fargo.

As you evaluate your upcoming product investments, how should we view the relationship between Family of Apps revenue and cost growth? Can you provide insights into that relationship? Additionally, regarding the elevated growth in G&A for Q1, did any one-time events contribute to that growth?

Addressing your second question first, the elevated G&A was primarily driven by legal-related expenses. We recognized some accruals in Q1 pertaining to ongoing legal matters, which will be detailed further in the 10-Q. Regarding the first part of your question about the long-term margin profile of Family of Apps, we aren’t issuing specific guidance on that; however, we have maintained a disciplined operational approach throughout 2023, which will continue moving forward. We are focusing heavily on judicious allocation of new resources and have built a strong foundation for this discipline, particularly since the Family of Apps is at an impressive scale.

Your next question comes from the line of Ross Sandler from Barclays.

Mark, you collaborated with Google and Bing to enable Meta AI organic search citations. Given this, do you believe Meta AI could potentially attract search advertising revenue in the long run, or do you foresee a model where you possibly implement a premium subscription tier once user engagement deepens? Additionally, you mentioned developing AI tools for businesses and creators. How do you envision the business model evolving as we progress towards utilizing customized AIs for creators like Taylor Swift for merchandise or ticket sales?

Regarding the partnerships with Google and Microsoft, we collaborate with them to integrate real-time information into Meta AI, which is useful. I think it's quite different from traditional search advertisements, and we aren't pursuing search ads specifically. Ultimately, this will likely constitute a different business model altogether. I believe there will be opportunities for advertisements and paid content interwoven into Meta AI interactions over time, as well as options for users to pay for enhanced models or increased computational resources. However, this is all still in the early stages of development. The most significant opportunity lies in our work around business messaging. This expands beyond the already-established efforts aimed at enhancing engagement and ad quality in our applications. With respect to business messaging, whether for creators or the over 100 million businesses on our platform, our aim is to simplify setting up AI engagement with their communities. For businesses, this means enabling sales and customer support interactions, while for creators, it could provide a fun and engaging element, as many creators use our platform to grow their businesses, whether they’re selling concert tickets or other products. Many of these creators aren’t leveraging advertising to its full potential. Similarly, the business messaging environments still remain relatively undervalued in terms of monetizable potential. Moreover, costs related to engaging audiences in messaging are still relatively high, but AI should significantly reduce these costs for businesses and creators. I think this presents a near-term opportunity, although it won't arrive within the next quarter or two, but it's not five years out either. Additionally, as Meta AI expands, I believe there will be monetization opportunities inherent in that scaling process as well, which we will develop gradually over time. Currently, our primary focus should be ensuring that hundreds of millions, if not billions, of people consider Meta AI as an integral part of their daily activities. This is the next milestone on our journey.

Krista, we have time for one last question.

And that question comes from the line of Ron Josey from Citi.

Mark, to revisit a previous question, you expressed that internal optimism has significantly increased due to ongoing improvements and innovations being made. We've experienced Meta AI for a few days; can you elaborate on how the $400 billion parameter model might transform the Meta experience or how you envision future developments in the upcoming months or years, particularly as messaging becomes a greater focus?

Yes. The forthcoming phase for several features is to tackle more complex tasks and develop more like agents rather than simple chatbots. By 'chatbot,' I mean a system that merely responds to your inquiries; it establishes a near 1:1 exchange. In contrast, an 'agent' will respond to your intent or goal and then proceed to execute multiple queries in the background to help meet that objective, whether it involves researching something online or selecting the right item to purchase. The potential complexity behind what users may be able to request from AI is significant. I believe that larger models—and future variations that may be smaller—can enable more fascinating interactions. For instance, the business use cases we discussed necessitate more than merely a sales or customer support chatbot fostering 1:1 interactions. When considering a business, there are broader objectives, such as delivering quality customer support and effectively marketing products tailored to consumer interests. The upcoming multiturn interactions will become vastly richer and more intricate, thanks to new reasoning and planning capabilities of emerging models. The opportunities here are substantial. Additionally, I believe we have effectively demonstrated our capacity to build leading models, so it is sensible to fully pursue this ambition, and we will. I envision this as a valuable long-term investment. I wanted to emphasize our focus and commitment to investing in this for the long term, as that aligns with our strategic approach.

Great. Thank you for joining us today. We appreciate your time, and we look forward to speaking with you again soon.

This concludes today's conference call. Thank you for your participation, and you may now disconnect.