Intelligent Financial APP learned that on October 31, Meta (META.US) held the 24Q3 earnings conference. The company's CEO stated that capital expenditures for the fourth quarter are expected to increase compared to the third quarter, partly due to increased server spending and increased data center capital expenditures. However, in terms of servers, the timing factor is at play as servers were delivered later in the third quarter. Therefore, cash will basically not flow out until the fourth quarter, when capital expenditures will become apparent. Given the general nature of capital expenditures, there will be some fluctuations in capital expenditures every quarter. Therefore, it is difficult to infer from any specific quarter. Overall, there has been a significant increase in infrastructure investment this year, with another significant increase expected in 2025.

Meta indicated that in the third quarter, the company added many new features, including providing account insights for businesses and creators to understand their later performance, the ability to save multiple drafts, continued commitment to integrating Thread with Fetaverse, basically focusing on expanding Thread's functionality over time, and responding to what users inform them they are interested in. Specifically in terms of monetization, the company expects Thread to not be a significant revenue driver in 2025. The company is currently only satisfied with the growth trajectory and emphasizes once again, focusing on launching features that the community deems valuable, striving to deepen growth and engagement.

The company's CEO pointed out that quick message delivery on WhatsApp monetization and the company's current status are key areas for the company, as there has been sustained development in this area. In particular, the growth of clicking on WhatsApp ads remains particularly robust. Therefore, the company will continue to focus on expanding the scale of WhatsApp ad clicks in markets with high WhatsApp adoption rates, such as Brazil. For example, in the United States, this will obviously be sooner, but there has been good growth in WhatsApp ad clicks and will continue to invest in expanding WhatsApp's scale and consumer adoption in the United States, which will create greater opportunities for the next steps.

Of course, the company is also doing a lot of work to improve the efficiency of information ad clicks and help advertisers optimize specific conversion events they are interested in. Another factor in WhatsApp revenue is paid messages, which continued to show strong growth momentum this quarter, actually remaining the main driver of growth for other revenue items in the app series, with a growth of 48% in the third quarter. We see a widespread strong growth in paid conversation volume, benefiting from both the growth in the number of enterprises adopting paid messages and the growth in conversation volume for each enterprise.

In terms of the growth of average ad prices, the group stated that driven by strong demand from advertisers, the average unit price increased by 11% year-on-year, partly due to improved ad performance over time. The growth rate of the average unit price slightly accelerated from 10% in the second quarter, partly because impressions grew lower in the third quarter. However, more broadly, when pricing growth and this metric, the year-on-year growth in reported prices per ad is influenced by many factors, including fluctuations and auction dynamics brought about by impression growth. One point the company is very focused on is input indicators, such as how much conversion rate is provided to advertisers? Over time, have they gained more value?

Q&A

Q: I would like to ask about MetAI. Could you help us understand some of the frequent types of interactions or queries you see with this product, and whether they have commercial intent? And then, over time, how do you consider building your own internal search product rather than partnering with other companies to handle these queries? Also, regarding the number of employees, because you mentioned a lot about the 25-year infrastructure investment. How should we consider relative personnel investment in 25 years to support all infrastructure, as opposed to what you have been doing in 2024? Thank you.

A(Susan, CFO): Your first question is about the recurring interactions we have seen as people use Meta AI. We see - first, I think as Mark mentioned, we are excited about the progress of Meta AI. Obviously, the development of Meta AI is still in its early stages, but it will continue to move towards the goal of becoming the most widely used artificial intelligence assistant globally by the end of the year, with its current monthly active users exceeding 0.5 billion.

People use it for a lot of things. Some common use cases we see include information gathering, helping with operational tasks, which is the biggest use case. But we also see people using it to delve into interests, find content in our services, generate images, which is another quite popular use case so far. What I want to say is, in the short term, our focus is to make Meta AI more and more valuable to people.

If we succeed, we believe that the scope of people using it will continue to expand, including more profitable queries over time. As for the second part of your question, Meta AI extracts information from the entire network, promptly addresses user queries, and provides the sources of these results from our search engine partners. We have integrated with Bing and Google, both of which deliver excellent search experiences. Like other companies, we also train our Gen AI models on publicly available online content and crawl web content for various purposes. Your second question is actually about how we are considering the number of employees in 2025.

We are still working on completing the budget for 2025. This is also part of why we are changing our forward-looking guidance methodology, providing guidance in the next call. But in our process, we are looking for opportunities to invest in our strategic priorities, which include monetization, infrastructure, reality labs, Gen AI, and our ongoing investments in regulation and compliance. When evaluating each opportunity, we focus on measurable ROI or strategic opportunities, depending on the area. We will continue to focus on streamlining operations in other areas to support the above work. So, we do not have specific information on the growth of employees in 2025, but this can give you a slight idea of our budgeting process.

Q: Comments on what you learned in the business planning process. Mark, I would like to better understand, as you consider your platform, product portfolio, and internal processes, what are the major sets of opportunities for applying artificial intelligence, since you sound quite optimistic about key learning outcomes and how they will continue to increase, and maybe even accelerate potential return cases. I just want a deeper insight into your main experiences in this process. Thank you.

A(Mark, CEO): I think the key here is that it seems to be widely applicable across various products. So, there are some areas that are part of our core business, from making feeds more relevant, making reels more relevant to making ads more relevant, helping advertisers create better ads, helping people create the content they want, assisting in our integrity operations and compliance, and the work we do in this area, all of this is important. In all these aspects of the core business, it is very valuable, and it will also bring new types of services, just like Meta AI that we didn't have before.

We didn't have products like Ray Band Meta before, and artificial intelligence will become a crucial part of all these things. There are other similar new products, such as Artificial Intelligence Studio (AI Studio). This year, our real focus is to launch Meta AI as our single assistant that people can ask any question to, but I think there are still many opportunities. I believe next year, in consumer and enterprise use cases, we will see more opportunities for people to interact with various different artificial intelligence agents. Consumers can use AI Studio, whether it's different creators, or different agents created for entertainment by people. Or in the business aspect, we also hope to continue making progress on this vision, allowing any small business or any company to establish an artificial intelligence agent by clicking a few mouse clicks, to help them provide customer service and sales services to customers worldwide, I think this is a huge opportunity. So, it's a very broad opportunity. I think there are many opportunities we see. Some long-term opportunities around Meta AI or AI Studio, may not necessarily bring huge profit opportunities in the next few years, but many things around participation and monetization in the core business, I believe will appear in the next few years. So, I think we are working hard to ensure that the right people are doing this work and making sure we have the right amount of investment for what we think are very very big opportunities.

Q: I would like to follow up on the situation with Meta AI. What I mean is, understanding how people are currently using this platform will definitely help us. Perhaps you can talk more about how, over time, with the introduction of agents, you truly expect use cases to not be limited to longer, more complex queries only? Then, Susan, regarding capital expenditures, I just want to understand more about your comments on the fourth quarter, it sounds like some payments have been pushed to the fourth quarter, guiding capital expenditures for this quarter to be $15 billion to $17 billion. Should we consider the run rate for 2025? Thank you.

A (Mark, CEO): I can answer questions about Meta AI, although I am intentionally not saying much about the new features and modes of MetAI that we are about to launch. However, as I mentioned in earlier comments, with each generation of products, I expect a large number of new features to be added. But I believe that this will be - this is also the part that excites me, and we will discuss this issue more when we are ready next year. However, I believe one of the trends we will see is that these models not only empower individual assistants in Meta AI but also drive growth between the AI studio and business agents. What I mean is, this year, if you look back at our situation a year ago, we started launching Meta AI this year. So far, we have successfully developed it and many people are using it. Over time, we also want to increase more depth in engagement and new use cases. Therefore, I think next year, our goal will be to work on popularizing these use cases, although achieving the depth of use and business results we want will take several years. So, we still have a lot of work to do, and I am excited to start discussing this issue early next year.

A (Susan, CFO): Your second question is about fourth-quarter capital expenditures. You know, fourth-quarter capital expenditures are expected to increase from the third quarter, partly due to increased spending on servers, followed by increased spending on data center capital. However, as for servers, the timing factors you mentioned come into play because we delivered servers later in the third quarter. Therefore, cash will basically flow out in the fourth quarter, when capital expenditures will become apparent. Given the general nature of capital expenditures, there will be fluctuations in capital expenditures each quarter. So, it's difficult to infer from any particular quarter. Overall, I'd say we're growing our infrastructure investments significantly this year and we expect significant growth again in 2025.

Q: This time I would like to ask a cost question. How do you internally use AI for productivity and efficiency, especially in the R&D department? How flexible is your employee count, Susan, when considering cost increases in other areas?

A (Susan, CFO): When it comes to using AI for productivity and efficiency, this is certainly something we are very excited about. I don't have any specific quantifiable data to share at this time. I think AI presents different efficiency opportunities that we have been monitoring. We can reduce costs and save time by improving internal productivity in areas such as coding.

For example, although it is still early days, we are seeing widespread adoption of our internal assistants and coding agents internally, and we are continuing to increase llama's efficiency in coding, which will make this use case more valuable to developers over time. Additionally, we hope that over time, we will be able to deploy these tools in many content moderation tasks to help us complete a large amount of content moderation work, thereby enhancing our efficiency and effectiveness. In many other areas of the company, we are exploring relatively early opportunities to use LLM-based tools to improve efficiency in various workflow types.

So, we are very excited about this. We have many teams focusing on this. From some small opportunities and G&A functions to larger opportunities that we hope will exist long-term in areas such as content management and coding productivity. As for your second question about employee count, you know, we really - to reiterate, our budget is still at a moderate level. So, we currently do not have much specific information to share.

But when we assess where there are good investment opportunities, we really consider opportunities with a very strict focus on return on investment. We carefully evaluate investment returns, opportunities for returns, likelihood of returns, and the overall increment of these investments when considering investing in core business. These are the issues we evaluate when assessing the return on investment that we believe can be achieved in the near term.

At the same time, we are also assessing opportunities in some medium-to-long-term strategic investment areas, including our efforts in the field of AI and supporting the infrastructure needed for AI, as well as our investments in reality Labs. These are all contents we are evaluating and the investment portfolio of things we believe can be done in 2025.

I have a few ideas, firstly, in which areas can we establish the greatest flexibility when considering infrastructure or personnel planning. Secondly, we place great emphasis throughout the entire company on improving efficiency and ensuring that we feel we are continuing to drive the development of the entire company, including areas where we expect to make more personnel investments, to think about how to be more efficient in 25 than in 24.

Q: Mark. You mentioned before that the higher the standardization of llama, the greater the improvement in Meta's core business. I was wondering if you could elaborate on this? Many developers are using Lama series models to build different things in the field of artificial intelligence. How are you leveraging this favorable condition to incubate new ideas internally within Meta? The second question you mentioned in a podcast after Meta Connect, assuming the scaling laws remain the same, we may need hundreds of billions of dollars in computing capital expenditure to achieve the goals around Gen AI. So I was wondering, given constraints such as energy, custom ASICs, or other factors, how quickly can you build up so much infrastructure? Can you provide us with more information on launching so much computing power at Meta? Thank you.

A (Mark, CEO): Yes, I can try to explain. What I mean is, for llama improvements, I would say there are several directions. There is the quality aspect, as well as the efficiency aspect. Many researchers and independent developers are working on llama, making improvements, and then releasing them, which is very easy for us because these jobs and showcased examples are all done by people on our stack, so we can easily incorporate them into llama and our meta products (such as Meta AI, AI Studio, or BusinessAI). Perhaps more important is efficiency and cost. I mean, these things are obviously very expensive. If someone finds a better way, if they can run 20% more efficiently, that would save us a lot of money. This is our experience in open computing, and this is why we initially place such importance on the open-source part, which is that we found a phenomenon in open computing that is contrary to intuition, which is that by releasing and sharing our computing architecture and designs, the industry has standardized it more. We also receive some suggestions that help us save costs, which is very valuable to us. One of the biggest costs here is chips and a lot of infrastructure. As Lama is increasingly adopted, companies such as NVIDIA and AMD have optimized their chips more to run Lama better, which is obviously beneficial to us. So, everyone using Lama can benefit from it, which in turn makes our products better, rather than building a model in the industry with no standardization on the island. These are some of the situations we see with llama, and I think that's why I believe it's a good business model for us to do this in an open way.

Regarding expanding infrastructure, what I mean is, when I talk about our team executing well, some are in order to, you know, provide more attractive products, some are to generate more revenue. On the infrastructure side, it is to quickly complete expenditures, right? So, I think what we've seen this year is that the infrastructure team has performed quite well. I think that's why in this year, we've been able to build more capacity. What I mean is, at the beginning of the year, we had a scope that we thought was possible. And what we were able to achieve, I think is more than we were hoping for or expecting at the beginning of the year. While this reflects increased spending, I am actually pleased that the team has done a good job of completing the tasks. I think this level of execution makes me more optimistic that we will be able to continue to develop at a good pace. But this is also part of the whole thing, the formula surrounding infrastructure development, which may not be the news investors want to hear about what we are developing in the short term. But I think the opportunities here are really huge. We will continue to invest heavily. I am proud of the team's excellent work, as they are building a lot of capabilities so that we can provide world-class models and world-class products.

Q: Mark, I would also like to ask a big-picture question, this time about Thread, which is now one of the core apps, becoming the next major social app with 0.275 billion MAU. I would like to hear your views on how this product will evolve over time, especially from a monetization perspective, and the next steps in terms of users. Then, Susan, the price has increased by 11% this quarter, I would like to understand the pricing dynamics of the platform. I think you mentioned that due to increased demand for ads and improvements in ad effectiveness, pricing is also increasing. So please help us understand this further. Thank you.

A (Susan, CFO): Your first question is about Thread. We are progressing well in this area. We will continue to launch more features and enhance our ranking stack. We are very satisfied with the continued growth of Thread users. We are increasing the number of users every quarter, and engagement is also steadily increasing. In the third quarter, user growth was particularly strong in major markets such as the USA, the USA, and Japan.

Yes. In the third quarter, we introduced many new features, including providing account insights for businesses and creators to understand their performance, the ability to save multiple drafts, continuing to fulfill the promise of integrating Thread with Fetaverse, essentially, we are very focused on continuing to expand Thread's features over time and responding to what users tell us they are interested in. Specifically in terms of monetization, we do not expect Thread to be a significant driver of revenue in 2025. We are merely satisfied with the growth trajectory and emphasize once again that we will truly focus on rolling out features that the community finds valuable, striving to deepen growth and engagement.

Your second question is about the average price growth of each ad. Driven by strong demand from advertisers, the average unit price increased by 11% year-on-year. Part of the reason for this growth is the improvement in ad performance over time. We see the growth in average unit price slightly accelerate from 10% in the second quarter, partly due to lower impression growth in the third quarter. However, more broadly, when we consider pricing growth and this metric, the year-on-year growth in reported prices per ad, you know, there are many factors at play, including fluctuations and auction dynamics resulting from impression growth. One key focus for us is the engagement metric. Do advertisers find value in what we provide in terms of conversion rates? Are they getting more value over time?

The blended reported price for each ad is complex because all these things are part of the mix. Our advertising clients have many different optimization objectives, each with very different values, making it challenging to compare apples to apples. However, we are closely monitoring the growth in conversion rates, which continues to outpace impression rate growth, and whether we are seeing healthy trends in action costs or per conversion costs, as we are doing. As long as we can continue to drive advertiser conversion rates, CPM should increase over time as we provide more value in terms of conversions for each impression, leading to higher impression value.

Q: I have a big-picture question similar to the ecosystem, which is when do you think we are going to see widespread adoption of third-party artificial intelligence applications, especially on the consumer side? I know we are seeing more and more agents on the enterprise side. But when can we see - how far out can we see - see consumer applications becoming pervasive in the field of artificial intelligence? How do you view - and how do Meta view itself as those critical - it used to be one of those key applications - in mobile internet and desktop internet, but now you also seem to be a player in the infrastructure space. So I would love to hear your thoughts. Thank you.

A (Mark, CEO): We are developing many consumer products. With Lama, I hope that app developers can also create many great products. I've mentioned Meta AI, AI Studio, and Business AI, I hope these will all become important parts of the consumer experience. Another part I haven't talked much about is, AI has the opportunity to help people create content, improve their feed experiences. Looking back at the big trends and pushes in the company's history, it started with the News Feed, right? So all updates basically come from content your friends post. Then we entered this era, where we also included creator content, and now a large part of the content on Instagram and Facebook is not from your friends. It might not even be from someone you directly follow, but maybe just recommended content from creators, and we can use algorithms to determine that this content is interesting, engaging, and valuable for you.

I believe we will add a whole new category of content, namely content generated by AI or summarized by AI, or existing content integrated in some way by AI. Over time, this will be very exciting for Facebook and Instagram, as well as for thread or other types of feed experiences. We have already started testing different things around this. I don't know if we know exactly what will be very effective. Some things are hopeful. I don't know if this will have a significant impact on the business in 25 years. But I think in the next few years, this will be an important trend and one of the important applications. But you will get, you will get many things that AI Studio, Business AI, and developers can do with Llama.

Q: Mark, Meta AI now seems to be able to become a network and provide conversational answers about anything (including current events). With over 10 million advertising clients and one of the best operating system products in your core business, do you plan to start testing commercial query advertising, to bring Meta AI closer to being a true answer engine, serving the billions of queries you have already seen? Then, Susan, one of the biggest obstacles we have encountered is RL losses, which were $16 billion last year and may exceed $20 billion this year. The question is, are we close to the peak of losses, or, which products do you think have the most potential in the coming years? Thank you.

A (Susan, CFO): Your first question is about plans to provide ads in commercial queries. I think I may have mentioned this question earlier. Now, our real focus is on how to make Meta AI provide as engaging and valuable experiences for consumers as possible. Over time, we believe the range of queries people use it for will expand, and I think over time, monetization opportunities will follow. But for now, what I want to say is, our real focus is on consumer experience, which is also a way we play when launching products globally, and before we focus on potential monetization scenarios, we will genuinely fine-tune the consumer experience.

The second part of your question is about Reality Labs. We currently do not disclose expectations beyond 2024. As we consider the budget process for 'Reality Labs' in 2025, we are also considering where we should focus our attention and energy. With the progress we have made in smart glasses and the strong consumer interest in smart glasses, we are once again very excited. So we are contemplating how to ensure our investments properly support the consumer momentum we are seeing. Overall, I think Reality Labs is clearly one of our long-term strategic priorities, and we hope it will be a significant investment area for us as we build ambitious product roadmaps there.

Q: First, this year we have encountered many unusual events that may drive the growth of advertising revenue, not only the U.S. election, but also elections in Europe and India, as well as major sports events like the World Cup and other events. Susan, when considering advertising revenue for next year and the future, will you mention the impact of these events that occur only every four or five years on advertising revenue? Secondly, can you discuss the monetization of WhatsApp and your current situation? It sounds like part of the business information. It has indeed been well integrated with other revenue sources. But please help us think about how far we are from optimizing WhatsApp's current level of monetization compared to two or three years from now. Thank you.

A (Susan, CFO): Mark. Your first question is about the revenue background for 2024, mentioning events that happen every four or five years. So I assume you are referring to the Olympics. Historically, we have not seen significant incremental contributions from events like the Olympics. We believe this year is no different. Therefore, looking ahead to the fourth quarter and next year, we generally expect growth to continue benefiting from the healthy global advertising demand we have seen. We believe our investments in improving advertising performance will continue to benefit our ad clients. However, obviously, there may be significant changes in the macro environment, which we try to reflect in our revenue guidance. But I'm not sure if there are many specific events that would have a significant impact on revenue in 2024.

Your second question is about WhatsApp monetization and our current status. Now, what I want to say is that quick clicking into messages is indeed our focus area. We have seen continued growth in this area. In particular, the growth of clicking on WhatsApp ads remains particularly strong. Therefore, we will continue to focus on expanding the scale of WhatsApp click ads in markets where there is a higher adoption rate, such as Brazil. For example, in the U.S., this is obviously earlier, but we have seen good growth in WhatsApp click ads and will continue to invest in expanding the scale and consumer adoption rate of WhatsApp in the U.S., which will create greater opportunities for the next steps. Of course, we are also doing a lot of work to improve the efficiency of message ad clicks and help advertisers optimize specific conversion events they care about. Another element of WhatsApp revenue is paid messages, which continued to show strong growth momentum this quarter, and still remains the main driver of growth for our suite of other revenue projects, growing by 48% in the third quarter. We see strong growth in paid conversation volume, benefiting from the growing number of enterprises adopting paid messages and the growth in conversation volume for each enterprise.