Meta: The surge in Q4 Capex has seasonal disturbances (3Q24 call notes)

The following is the summary of the Q3 2024 earnings conference call for $Meta Platforms(META.US) . For the financial report interpretation, please refer to《 Meta: Completely Becoming an "AI Maniac," Can High Growth Withstand High Investment? 》

(Article link: )

1. Core Financial Report Information Review:

2. Detailed Content of the Financial Report Conference Call

2.1. Key Information from Executives:

Business Progress

1）Overall Performance

The number of daily users of Meta applications worldwide exceeds 3.2 billion, and Meta AI and Llama are being rapidly adopted by the industry.

2）Application Product Side

a. WhatsApp: The United States is one of the fastest-growing countries, with global daily call volume exceeding 2 billion.

b. Facebook: The growth trend of young users continues, especially in the United States. A teen account has been launched with protective measures to limit their information exchange and content browsing.

c. Threads: The community now has nearly 275 million active users monthly, with over 1 million new users daily**, and user engagement has significantly increased.

d. Advertising: Over 1 million advertisers have used generative AI tools, creating more than 15 million ads just last month, with a 7% increase in conversion rates.

3) Meta AI

a. Meta AI Improvements: The AI-driven content recommendation system has brought about an 8% increase in engagement.

b. Llama Model: Llama 3.2 has been released, including small models for devices and open-source multimodal models. Llama has also been applied within the U.S. government. The Llama 4 model is in the development stage, featuring stronger reasoning capabilities and faster processing speeds.

c. Investment Planning: Plans to continue investing in AI infrastructure to accelerate core business, believing these businesses will have strong ROI in the future.

4）Reality Labs and Wearable Devices

a. Ray-Ban Meta Glasses: Combining the functionalities of Meta AI, they can provide real-time suggestions, language translation, and can "see" the scenes that users seeb. AR Glasses: The Connect conference showcased the Orion Holographic AR Glasses, aimed at seamlessly integrating the physical and digital worlds.

c. Mixed Reality Headset: The latest Quest 3S mixed reality headset has been released at a price of $300, positioned to be more affordable.

5）Business Outlook

a. User Experience and Interaction: Meta is committed to improving user experience within applications and investing in long-term projects to enhance user interaction, particularly in content recommendations.

- Short-term Strategy: Facebook will launch a globally unified video player and immersive experience system, expected to complete this global rollout by early 2025; Instagram continues to promote original content.

- Long-term Strategy: Develop new content recommendation models to improve recommendation accuracy and engineering efficiency.

b. MetaAI: Usage is expanding in more countries and languages, with voice interaction features fully launched in the United States, Australia, Canada, and New Zealand. Based on the Llama 3.2 multimodal model, it supports user photo uploads, image content editing, and more.

c. Threads: The number of Threads users and interaction depth continues to grow, with more topic tracking features to be launched in the future.

6）Monetization Efficiency

a. Advertising Optimization: Continuing to optimize ad placement and frequency, especially for video ads.

b. Advertising Performance Improvement: New learning and modeling methods have been adopted to achieve breakthroughs in user behavior sequence analysis within the advertising system, increasing conversion rates by 2%-4%; updates to the ad ranking and optimization model were made in the third quarter, expected to enhance Meta attribution conversions for advertisers.

c. Advantage+ Solutions: The generative AI-driven advertising creative tool has seen strong advertiser retention, with video generation and image animation features in testing set to be fully launched early next year.

2. Financial Performance

1) Overall Overview

a. Revenue: Total revenue for the third quarter of 2024 was $40.6 billion, a year-on-year increase of 19% (20% growth at constant currency).

b. Expenses: Total expenses were $23.2 billion, a year-on-year increase of 14%. The main growth came from infrastructure costs and R&D personnel costs.

c. Net Income: Net profit was $15.7 billion, with earnings per share of $6.03.

d. Operating Profit Margin: Operating profit was $17.4 billion, with an operating profit margin of 43%.

e. Free Cash Flow and Capital Expenditures: Free cash flow reached $15.5 billion; capital expenditures were $9.2 billion.

2）Family of Apps

a. Revenue: Third-quarter revenue was $40.3 billion, a year-on-year increase of 19%.

b. Advertising Segment:- Advertising revenue was $39.9 billion, up 19% (up 20% at constant currency)

- Revenue growth breakdown (by industry): Online shopping, healthcare, entertainment, and media are the main sources of advertising revenue growth.

- User revenue growth breakdown (by region): The regions with the highest user revenue growth are Other Global Regions (23%) and Europe (21%), followed by Asia-Pacific (18%) and North America (16%).

- Advertiser revenue growth breakdown (by region): North America and Europe both at 21%. Other regions grew by 17%, while Asia-Pacific grew by 15%, slowing from the 28% growth rate in the second quarter.

- Ad impressions increased by 7%, with the average price per impression rising by 11%.

c. Other segments: Family of Apps other revenue grew by 48%, reaching $434 million, primarily from business messaging revenue on the WhatsApp business platform.

d. Expenses: Family of Apps expenses were $18.5 billion, up 13% year-on-year, accounting for 80% of total expenses, mainly due to rising infrastructure and labor costs.

e. Operating profit: Family of Apps operating profit was $21.8 billion, with an operating margin of 54%.

3) Reality Labs business

a. Revenue: Third-quarter revenue was $270 million, up 29% year-on-year, primarily driven by hardware sales.

b. Expenses: Reality Labs expenses were $4.7 billion, up 19% year-on-year, mainly due to increased labor and infrastructure costs.

c. Operating loss: Reality Labs operating loss was $4.4 billion.

4) Capital allocation strategy

a. Ensure returns in core business and long-term opportunities through investments in infrastructure and talent.

b. Plans to expand the infrastructure scale for smart glasses and generative AI.

5) Q4 and full-year guidance

a. Revenue guidance: Total revenue for Q4 2024 is expected to be between $45 billion and $48 billion.

b. Expense guidance: Total expenses for the full year 2024 are expected to be between $96 billion and $98 billion.

c. Capital expenditures: Full-year capital expenditures are expected to be between $38 billion and $40 billion, with significant growth anticipated in capital expenditures in 2025.

d. Tax rate: The tax rate for Q4 is expected to be in the low teens percentage.

e. Regulatory risks: Closely monitoring legal and regulatory challenges, including those in the EU and the US, which could have a significant impact on business and finances.

2.2 Q&A Analyst Q&A

Q: Regarding Meta AI, can you help us understand the main types of interactions or queries that the product currently supports, and whether there are any commercial intentions involved? Additionally, in the future, do you prefer to build an internal search engine functionality yourself, or continue collaborating with other companies?

A: Users utilize Meta AI to accomplish many tasks. Some of the high-frequency use cases we have observed include information queries, obtaining deeper insights about content of interest, searching for content on our services, and image generation, which is currently a very popular use case. Over time, we are also adding more query types with commercial potential.

Meta AI extracts information from web content to answer users' real-time questions, and provides the sources of these results through our search engine partners (Bing and Google). The company also trains generative AI models using publicly available web content and crawls web content for various purposes.

Q: How do you view the investment in personnel for 2025 in support of infrastructure development compared to 2024?

A: Regarding personnel allocation for 2025, the company is still in the budgeting phase. When considering future investment areas, we focus on strategic priorities in monetization, infrastructure, Reality Labs, generative AI, and compliance, and evaluate based on the return on investment or strategic opportunities in each area. The company is also streamlining operations in other areas to support these investments, but there are currently no specific details on personnel growth for 2025.

Q: Can you elaborate on the biggest opportunity areas for applying AI in platforms, product portfolios, and internal processes? Are you optimistic about these key learnings and the potential returns from AI applications, and can you provide more details?

A: The application of AI is very broad. In our core business, enhancing the relevance of Feed and Reels, optimizing ad performance, supporting content creation, and strengthening compliance and operational work are all areas that will benefit from AI. AI not only holds significant value for our core business but will also drive the emergence of entirely new service types, such as Meta AI and Ray-Ban Meta glasses, which will incorporate AI as a key component.

Additionally, we have launched new products like AI Studio. This year, we focused on launching Meta AI as a single assistant capable of answering any question. In the coming year, we will see rapid growth in use cases in both consumer and business sectors. For example, consumers can interact with different AI agents through AI Studio, covering content created by entertainment content creators and other users. In the business sector, we hope that every small business or company can set up an AI agent easily to provide customer service and sales support to global customers, which will be a huge opportunity.

We see many development opportunities, some of which are long-term, such as Meta AI and AI Studio, which may not generate significant returns in the near term. However, in core business areas, such as user engagement and monetization rates, AI will bring significant returns in the coming yearsWe ensure that we invest the right personnel and resources to seize this huge opportunity.

Q: Regarding Meta AI, with the introduction of agent functionality, how do you anticipate the future use cases will expand?

A: Currently, we will not disclose too many details about the new features and models that Llama 4 will bring to Meta AI.

However, it is foreseeable that these models will not only support a single assistant for Meta AI but will also further expand into AI Studio and business agents. This year, we have achieved significant success in promoting Meta AI, with increases in both user numbers and usage depth, and we will continue to add new application scenarios in the future. For AI Studio and business AI, they are currently roughly at the same stage as Meta AI was a year ago. Our goal for next year is to have these applications used more widely, although achieving the ideal depth of use and business effects will require several years of development.

Q: Regarding capital expenditures, some payments have been postponed to the fourth quarter, and capital expenditures for the fourth quarter are expected to be between $15 billion and $17 billion. Will this level also continue into 2025?

A: Capital expenditures in the fourth quarter have increased compared to the third quarter, mainly due to increased server spending, while capital expenditures for data centers have increased less. Server spending is affected by delivery times, as some servers were delivered at the end of the third quarter, and payments will be made in the fourth quarter, so capital expenditures will be reflected in the fourth quarter. Additionally, capital expenditures typically have some quarterly volatility, making it difficult to extrapolate from a single quarter's data. Overall, our investment in infrastructure has grown significantly this year, and we expect it to continue to grow substantially in 2025.

Q: How does the company internally utilize AI to enhance employee productivity? In particular, has there been a significant improvement in productivity within the R&D teams? Also, I would like to ask again about the flexibility of employee numbers in light of cost increases in other sectors.

A: First, regarding the application of AI in enhancing employee productivity, this is an area we take very seriously. Currently, there are no specific quantifiable results to share, but we are already focusing on how to improve internal efficiency through AI to reduce costs. For example, the internal application of AI in coding and other areas has already shown initial results, and the application of Llama in code writing will gradually enhance its value to developers. We also hope to apply these tools in content review in the future, making this substantial workload more efficient and precise.

In addition, other departments in the company are exploring how to use tools based on large language models to enhance the efficiency of different workflows. Currently, these applications are still in the early stages, but we have already seen small-scale efficiency improvements in G&A (General and Administrative) functions, and we hope to achieve greater improvements in areas such as content review and coding productivity in the future.

Regarding the question of employee numbers, the budget is still being formulated, so there is no definitive conclusion yet. However, in investment decisions, we will focus on ROI-driven positions, strictly evaluating their return potential and feasibility to achieve short-term investment returnsAt the same time, we are also considering opportunities in the medium to long-term strategic areas, including generative AI and the infrastructure that supports this technology, as well as investments in Reality Labs.

Overall, we will consider flexibility as a factor in our 2025 plans, especially in terms of infrastructure and employee allocation. In addition, the company is also promoting efficiency improvements to ensure that all departments, including those that may need to increase staff, are thinking about how to operate more efficiently in 2025 compared to 2024.

Q: The more standardized the Llama model is, the more improvements can feedback into Meta's core business. Can you elaborate on this point? The Llama series models are being used by many developers to build different AI applications. How does Meta leverage this perspective to incubate new ideas internally?

A: The improvements regarding Llama mainly reflect in quality and efficiency. Many researchers and independent developers are conducting research and optimization based on Llama and publicly sharing these results, which allows us to easily integrate these improvements back into Llama and apply them to Meta AI, AI Studio, or Business AI and other Meta products.

Moreover, increasing efficiency and reducing costs are also crucial. The operating costs are quite high, and if we can improve efficiency by 20%, it will save us a significant amount of expenses. The open-source architecture and design not only bring cost savings but also promote industry standardization, similar to our experiences in Open Compute.

For example, with the widespread adoption of Llama, chip manufacturers like NVIDIA and AMD are also optimizing their chips to better run Llama, which benefits everyone using Llama, including Meta's own products.

Q: It was previously mentioned that achieving the goals of generative AI may require hundreds of billions of dollars in computing capital expenditures. Given the current constraints of energy, custom ASICs, etc., how quickly can Meta realize the launch of this infrastructure? Can you further discuss the speed and challenges of achieving this computing capacity expansion?

A: In terms of infrastructure expansion, our team has been very effective this year, allowing us to build more computing capacity than initially expected at the beginning of the year. While this reflects higher expenditures, I am very satisfied with the team's performance. Although this investment may increase costs in the short term, due to the tremendous opportunities presented by AI, we will continue to invest heavily in infrastructure to ensure the provision of world-class models and products.

Q: Threads is gradually evolving into the next important social application, with monthly active users reaching 275 million. What are your views on the future evolution of this product, especially regarding monetization and the next steps for user growth?

A: Regarding Threads, we have made good progress, continuously launching new features and improving the ranking system, with steady increases in user growth and interaction depth. User growth in the third quarter was particularly strong in core markets such as the United States, Taiwan, and Japan. New features include account analytics for businesses and creators, support for saving multiple drafts, and continuing to advance the integration of Threads with the metaverse.In terms of monetization, we do not expect Threads to become a significant source of revenue by 2025 , currently focusing on community-recognized features to deepen user growth and interaction.

Q: Advertising prices increased by 11% in the quarter; can you further explain the pricing dynamics of the platform? Is the price increase related to increased advertising demand and improved advertising performance, and can you elaborate on this?

A: Regarding the increase in advertising prices, the average price per ad grew by 11% year-on-year, primarily driven by increased advertising demand and improved ad effectiveness. The CPM growth in the third quarter was slightly higher than the 10% in the second quarter, partly due to lower growth in impressions. Overall, price growth is influenced by auction dynamics and fluctuations in impressions. We focus on input metrics, such as advertisers' conversion rates and cost-effectiveness. Although the metrics for advertising prices are complex, as long as we can improve ad conversion effectiveness, CPM will increase accordingly, as the conversion value per impression will be higher.

Q: In the consumer space, how far are we from the widespread adoption of third-party AI applications? We are already seeing more AI agents and applications emerging in the enterprise sector, but when do you expect widespread adoption in the consumer space? Additionally, how does Meta view its role in this ecosystem?

A: We are developing multiple consumer-facing products, and with Llama, we expect application developers to be able to build many high-quality AI applications. Meta AI, AI Studio, and business AI will become important parts of the consumer experience. Another important area is using AI to help users create content to optimize the information flow experience.

Looking back at the development of the company's information flow, the initial content mainly came from users' friends, then we introduced creator content, and now a large portion of content on Instagram and Facebook does not come from users' friends or even those they directly follow, but rather from creator content recommended by our algorithms.

In the future, we will introduce a brand new category of content, namely AI-generated content, AI-summarized content, or existing content organized by AI, bringing a richer experience to platforms like Facebook, Instagram, and even Threads.

We are currently testing different methods of AI content generation; while it is still uncertain which will be the most effective, I believe this is one of the important development trends in the coming years. At the same time, Meta AI Studio, business AI, and developer applications based on Llama 2 will bring more possibilities.

Q: Meta AI can now access information online, providing conversational responses covering a wide range of content, including real-time events. Are there plans to test advertising in commercial queries and gradually develop Meta AI into a real answer engine to meet billions of query demands?

A: Our current focus is on enhancing the user experience of Meta AI, making it more attractive and valuable to consumers. In the future, as user queries expand, we believe monetization opportunities will arise, but the primary goal at this stage is to optimize user experience, which is also our consistent product strategy: prioritizing user experience before focusing on monetization.**

Q: Regarding the ongoing losses at Reality Labs, which amounted to approximately $16 billion last year and may exceed $20 billion this year. Is Reality Labs close to its peak loss, or which products in this field have the greatest potential in the coming years?

A: Regarding Reality Labs, we currently have not shared expectations beyond 2024. In our budget planning for 2025, we will focus on investment directions, particularly in smart glasses. We are excited about the progress in smart glasses and the strong interest from users, so we hope to appropriately increase our investment in this area. Overall, Reality Labs is a long-term strategic focus for us, and we will continue to invest heavily in achieving an ambitious product roadmap.

Q: When considering the comparison base for next year and the future, how do you think major events that occur every four to five years (such as elections in the U.S., Europe, and India, as well as important sporting events like the World Cup) will impact advertising revenue?

A: Events that occur every four to five years, such as the Olympics, have historically shown that such events do not significantly contribute to incremental revenue. This year has been largely the same. Therefore, when looking ahead to the fourth quarter and next year, we primarily expect growth to continue benefiting from strong global advertising demand. Our investments in improving advertising performance will continue to bring value to advertisers.

Despite various macroeconomic factors, we will consider them within the revenue guidance range, but we believe there are no individual events in 2024 that will particularly impact revenue.

Q: WhatsApp is currently making a positive contribution to revenue growth; can you elaborate on the monetization of WhatsApp? What is its potential in the next two to three years?

A: Regarding the monetization of WhatsApp, Click-to-Message is currently the main focus area, especially with strong growth in Click-to-WhatsApp advertising. We are expanding Click-to-WhatsApp ads into more markets with a strong WhatsApp user base, such as Brazil. Although the U.S. market is relatively new, we have also made good progress in Click-to-WhatsApp advertising and WhatsApp user growth, and we will continue to invest to expand opportunities. Additionally, we are committed to enhancing the effectiveness of Click-to-Messaging ads to help advertisers optimize specific conversion events that matter to them.

Another part of the revenue comes from WhatsApp's paid messaging service. This area has continued to grow strongly this quarter and is a major driver of growth in other revenue lines for family applications, with a year-on-year increase of 48% in the third quarter. This growth comes from more businesses adopting paid messaging services and an increase in the volume of conversations for individual businesses.

Risk disclosure and statement for this article: Dolphin Investment Research Disclaimer and General Disclosure