Meta Platforms: Focusing on AI Empowerment, Not Just Commercialization (2Q23 Earnings Call Summary)

The following is the summary of $ Meta Platforms(META.US) 23Q2 financial report call meeting. The financial report interpretation can review the " TikTok loss of power, meta Total Rebirth ''1. Management Overview It was a good quarter and we saw strong growth in our product and business. More than 3.8 billion people now use at least one of our apps every month, Facebook currently have more than 3 billion MAUs, and DAUs continue to grow worldwide, including the US and Canada. In addition to the good performance of our core products, I think we see the most exciting development trend I 've seen in a while. We have made continuous progress on Threads, Reels, Llama2, and some AI products in development. Quest 3 will also be released this fall. We are now working hard to execute all of this, and I am also pleased to see the decisions and investments we have made begin to take effect. In Threads, simply put, I'm fairly optimistic about our trajectory. We have seen unprecedented growth from the beginning, and more importantly, we have seen more people coming back every day than I expected. After that, we'll focus on growing the community to the scale we think is possible. Only then will we be committed to profitability. We 've run it many times before on Facebook, Instagram, WhatsApp, Stories, Reels, etc., and it's a great start to what we were hoping for, so I'm very pleased with the path we're on. One thing I 'd like to mention about our Year of Efficiency is that we 've seen examples of how our lean organization and some of the organizational changes we 've made have been able to build higher quality products faster, and Threads are probably the biggest examples so far. The efficiency year has two different goals: 1) to become a stronger technology company; 2) and to improve our financial performance. This allows us to actively invest in a longer-term roadmap. Now that we have completed massive layoffs, **the rest of 2023 will be about creating stability for employees, removing obstacles that are holding us back, introducing new AI tools to accelerate our growth, and so on * *. In the next few months, even though our financial situation has improved. We expect to continue hiring in key areas, but growth in new hires will be relatively low, meaning that as part of this year's layoffs, many teams are choosing to lay off employees rather than hiring people with different skills, so many hires will be delayed until 2024. The other major budget point we're looking at is. Since we don't know how fast our new AI products will grow, we may not have a clear grasp of this until **later this year * *. On our product roadmap, and I 've said on many calls, the two big waves of technology we're going through are the near-term AI and the long-term meta-universe. 1. Products and Business AI: Our years of investment in artificial intelligence have paid off in our ranking and recommendation system, and we are constantly improving user stickiness and commercialization. AI-recommended content from accounts that users aren't following is now the fastest-growing content category in the Facebook. Since the introduction of artificial intelligence recommendation systems, they have driven 7% of the total user time growth. This improves the user experience because the user can discover things he hasn't encountered before. In addition to Reels, AI drives our bottom line through our automated advertising product **Meta Advantage +. Almost all of our advertisers are using at least one of our AI-powered products. We also deployed Meta Lattice, a new model architecture that can learn to predict the performance of ads through various data sets and optimization goals. We also introduced AI Sandbox, a testbed for AI generation tools such as automatic text changes, background generation, and image display. Business Messaging: Business Messaging is another key part of our profit strategy, which we recently announced, without Facebook accounts. That's quite a change, especially in countries where WhatsApp are often the first step in bringing business online. Paid Messaging appeared earlier, but it also showed good adoption rates, with the number of companies using our Paid Messaging products doubling year-over-year. Speaking of instant messaging, we started rolling out Channels feature on the WhatsApp last month, which is an easy, reliable and private way to receive important updates from individuals and organizations. I am very happy to see more people try it. We will also try to bring the product to more countries and regions this year. LLM Big Model: In addition to the recommendation and ranking system in our products, we are also building a leading basic model to support a new generation of artificial intelligence products., and make it available for research and commercial use. We have a long history of open source infrastructure and AI work-from PyTorch (the leading machine learning framework) to Segment models like Anything, ImageBind, and Dino, to infrastructure as part of the Open Compute project. We 've found that open-sourcing our work allows entire industries, including ourselves, to benefit from innovation. And these are usually security improvements, because open source software will be more scrutinized and more people can find and determine the fix for the problem. These improvements also often come in the form of efficiency gains, which should enable us and others to run these models with reduced infrastructure investment in the future. I'm also looking forward to seeing the community's improvements to the Llama2 model. We are also using Llama ourselves to develop some new products that will serve our work, and I will share more details later this year, but you can imagine there are many ways that AI can help people-connect and express themselves in our applications, make sharing content easier, provide more interesting creative tools, act as assistants to help users interact with businesses and creators, and so on. Reality Labs: As our investments in artificial intelligence continue, we will also continue to work towards the vision of a meta-universe. Over the years, we have been working on these two priorities in parallel in a number of ways, and the two areas are overlapping and complementary. The new MR is able to seamlessly blend your physical and virtual worlds by intelligently understanding the physical space around you. We continue to break new ground in MR and our Quest Pro headsets, and the Quest 3 takes it to the next level. Of course, other companies in the industry are also working to bring MR technology to market, but the Quest 3 will be the first mainstream, available device,. Meta-universe content and software will also continue to merge. For Horizon, our team is now focused on improving retention, and we're making good progress on that. We 've also made significant improvements in our avatar, which will serve as a bridge between our mobile apps and VR/MR experiences. We will share more about our work on the metauniverse and AI at the Connect conference on September 27 at our headquarters. Now in addition to artificial intelligence, the other major wave of technology we're looking at is the meta-universe. For many years, we have focused on artificial intelligence and the meta-universe, and we will continue to focus on both-these two fields are also related. Breakthroughs in computer vision enabled us to launch the first standalone VR device. Mixed reality is built on a set of AI technologies for understanding the physical world and fusing it with digital objects. Being able to generate the world programmatically is important for delivering engaging experiences at scale. Our vision for AR glasses involves an AI-centric operating system that we think will be the foundation for the next generation of computing. Meta-cosmic technology will also help provide artificial intelligence. For example, incarnating AI agents will take advantage of the deep investments we 've made in avatars over the past few years. Building a meta-universe is a long-term project, but its basic principles remain the same, and we are still committed to it. Now in the short term, we 've reached some milestones that I think are worth mentioning. **Since last year, more than a billion meta-universe virtual avatars have now been created. The number of games in the quest store with revenue of at least $25 million doubled. More than half of the tasks daily active users now use their devices for more than an hour. The next milestone is that we are preparing to launch the next generation of consumer virtual and mixed reality devices later this year. **At this point in about 3 years, we launched the Quest 2. It's a big step forward for VR and I'm really excited to show the world all the improvements and new technologies we 've developed since then at a price that can be used by a lot of people. Two main factors drive our revenue performance-**our ability to provide an engaging experience for the community and our ability to commercialize this user stickiness * *. (1) User experience stickiness First of all, the overall user participation of Facebook and Instagram is still very high, and Reels continues to grow and promote the increase of user stickiness. On Facebook Feed in particular, the amount of content recommended by accounts that users don't follow has increased significantly over the past year, while user engagement has also increased-a sign that people are gaining added value from content that they don't follow. Looking ahead, we are optimistic that we have the ability to further enhance this value by leveraging advanced AI technologies to improve recommendations. In addition to increasing the value people receive in our family of apps, we are investing in new experiences for the future. We're building the infrastructure to support new AI-driven products coming online on our service, which will give people more tools to express themselves and connect. Since the Threads release of our new standalone app, we have been pleased with its initial response, and our focus is now on further developing it into a product that will be valuable to many people over time.(2) Another driver of the shift from commercialization capacity to revenue-increased monetization capacity. **We focus on improving the commercialization efficiency of products that are currently less profitable, such * * * *. **Reels We have made good progress on earnings,. We remain focused on further reducing the negative impact of Reels on revenue and closing the monetization efficiency gap with the company's more mature products. However, instant messaging: In the field of instant messaging, billions of people and millions of businesses use our messaging service to connect every day. We see significant opportunities to build tools and capabilities for businesses to help facilitate these interactions, and WhatsApp paid messaging solutions are making early but promising progress. In terms of our work on providing advertisers with measurable performance, it focuses on two main areas: artificial intelligence and onsite conversions. We are using artificial intelligence to shift our system to use fewer and larger models, allowing us to use learning across product functions and improve faster and more efficiently., The penetration of these products is increasing. We're seeing this effort translate to advertisers, and conversion growth remained strong in the second quarter. In terms of driving conversions, we continue to see that we also recently started testing the ability to buy Click-to-WhatsApp ads directly from WhatsApp commerce apps, and now have more than 0.2 billion users per month. Going forward, we will focus on enabling businesses to further optimize conversions in our messaging applications. Investment philosophy: We hope to establish discipline and habits during the year. At the same time, we remain focused on investing in the big opportunities ahead. Part of the way to support these initiatives is to prioritize them with other areas of work and transfer resources. However, **in some cases they will require incremental investment. This is especially true in areas where we see the most important opportunities, including artificial intelligence and the meta-universe * *. As I mentioned last quarter, we will also continue to focus on moderately developing our capital structure over time. We are pleased to have a second bond offering in May and expect future debt to increase at a modest pace as **we are working to improve our overall cost of capital while maintaining a positive or neutral net cash balance * *. In addition, we will continue to monitor the current regulatory environment. Regarding the EU-US data transfer aspect, we have seen a positive development in the European Commission's adoption of the final adequacy decision, which allows us to continue to provide services in Europe, which is good news. But overall, we continue to see increasing legal and regulatory headwinds in the EU and the US, which could significantly impact our business and financial results. 3.3Q23 Guidance: Revenue: We expect total revenue to be between $32 billion and $34.5 billion in Q3 2023. EXPENSES: Due to legal-related expenses recorded in the second quarter, we expect total expenses for the full year 2023 to be in the range of $880-$91 billion, an increase from the previous $860-$90 billion, and this guidance includes approximately $4 billion in restructuring costs related to facility consolidation costs, severance costs and other personnel costs. We expect Reality Labs' operating loss to increase YoY in 2023. While we do not currently provide a quantitative outlook beyond 2023, we expect several factors to be drivers of total fee growth in 2024 as we continue to invest in our most attractive areas. 1) We expect infrastructure-related costs to rise next year. Given our increased capital investments in recent years, we expect depreciation expense to increase more in 2024 than in 2023. We also expect higher operating costs from operating a larger infrastructure. 2) As our employee composition evolves into more costly technical roles, we expect compensation expenses to grow. 3) For Reality Labs, due to our ongoing AR/VR product development work and our investment in further expanding the ecosystem, we expect operating losses to increase year on year. CAPEX: We expect capital expenditures to be between $27 billion and $30 billion, down from our previous estimate of $30 billion to $33 billion. The downward revision of the forecast is due to cost savings, **rather than a reduction in the overall investment plan * *. Looking ahead, while we will continue to refine our plans this year, we currently anticipate that, driven by investments in data centers and servers, particularly in supporting our AI efforts, we will continue to expand our efforts. TAXES: If there are no changes to the U.S. tax code, we expect tax rates for the rest of the year to be similar to those in the second quarter of 2023. * * 2. Q & A **** Q1: Specific Use Cases or Advertiser Products for AI to Participate in Layout? What technical obstacles does AI face in the layout of the whole ecology? * * A1: At the upcoming Connect conference in September, we will announce the roadmap of our current development and the contents of some products. Our LLaMA2 model will underpin many of the new things we are building. In general, we plan to build three basic categories of products and technologies with generative AI: Different types of AI agents. 2. AI-driven generation functions, such as advertisers can help them run advertisements without having to provide many ideas or have suitable advertising images. A broad focus on internal productivity and efficiency, helping engineers write code faster, helping people understand the company's overall knowledge base internally, and more. One of the things that we think differently about this issue than some others in the industry is that we don't think there will be a single AI that people will interact with because people encounter these different entities every day, so **I think AI is focused on helping people connect in terms of expression and creativity, facilitating connections, and helping creators nurture their communities * *. The way to have the fastest direct business cycle is to help people interact with the enterprise. It will take time to achieve this, but this will **alleviate Meta's biggest problem with messaging monetization * *. Since the staff end is very human-intensive, with the help of AI agents, business messaging can spread around the world, just like the kind of success Meta has seen in Thailand or Vietnam. The company cannot predict the pace of new product development. When we consider the number of AI capital expenditures, we don't know how fast these will scale * *. We hope to have sufficient capacity to prevent a rapid expansion of capital expenditure. But because they are a brand new thing, there is not much precedent for such things. It's really hard to predict. Q2: How do you see the expansion of the developer community growing around LLaMA open source? Can LLaMA open source provide a way for companies to convert to B/To C? A2:LLaMA is an open source project, which is a bit different from building a developer platform. We do a lot of work on core infrastructure work, the design of servers and data centers, and basic infrastructure and artificial intelligence. We think it's good to have the community standardize what we're doing, and it helps with hiring, because a lot of the best people want to come and work where these things are being built, and they're used to the tools. Open source AI has come under more scrutiny, so it is ultimately more secure. Any improvements that people make to strengthen it in the community will be able to be incorporated into our work, making it easy to apply and expand across the industry. The company also expects efficiency gains. From the original LLaMA, some improvements have also occurred around quantification and being able to run models more efficiently. We can get a lot of help from the community to make them run more efficiently. This will not only help individuals run on laptops, but also hopefully translate into our infrastructure and save us considerable costs. We work with Microsoft because we don't have a public cloud product (not an area where we want to get involved), whereas Microsoft has a cloud product and other companies have cloud products. But when the open source license is open, **but in the long run, there may be some uncertainties. **** We do not expect LLaMA series products to generate separate revenue. **** Q3: Incentives for the loss of the VR sector is still increasing? **A3: The big event next year is the release of Quest 3, and it will take a lot of expenses to bring it to market. Our leadership in AR, VR and more gives us more confidence that the company is on the right path. The layout of the company's Reality Lab business is long-term and diversified. We will evaluate the return on investment from a long-term perspective, so **in the short term, our focus is to increase the adoption rate of existing products and continue to learn more about requirements and use cases * *. Significant investment will be around basic research and development to clear technical barriers, which will make subsequent equipment smaller, cheaper, and so on. **But the company's high hopes for Reality Lab have not changed, which remains an important long-term opportunity for the company. **** Q4: What is the company's strategic arrangement for this year and what is the focus? **A4: In recent years, the company's priorities have been consistent. Although the company's investment in artificial intelligence received considerable criticism last year, it proved to be a fruitful investment that drove results in the short term. Reels is not a new product, and Threads is not a huge project, and the company has a relatively small team in this area. Two things exceeded the company's expectations this year:, and created a huge opportunity. In the future, the company will continue to focus on AI and meta-universe. AI aspects include providing ranking and recommendation systems for core applications, as well as all advertising content seen in Facebook and Instagram. AI also supports companies in building their security systems. But on the other hand, although other companies are also doing related things, the company is confident that it will be the leader in (the meta-universe). **All in all, the company will continue to focus on AI and the meta-universe. **** Q5: Reasons for Q2 revenue growth? Did the adoption of "Advantage" drive the growth of the company's advertising business? **A5: Several factors drive Q2 revenue growth: 1. **Q1 from a low base * *. In Q1 the company completely experienced the Ukrainian war, service in Russia was forced to suspend. 2. 2Q23 The company ushered in an increase in supply and an improvement in advertising performance, including the company's improved Reels realization efficiency in response to the Reels revenue plateau * *. 3. 2Q23 The company faces **less foreign exchange resistance * *. "Advantage" is one of the many AI advertising products on the market. It has been widely used in e-commerce and retail verticals and has been successful. In particular, the company has seen the good attraction of "Advantage" in CPG and other vertical categories.The company will continue to roll out more features to make it easier for advertisers to adopt Advantage + for marketing and measure their performance benefits. The company is currently working on a "Advantage +" shopping campaign and a broader "Advantage +" portfolio to help advertisers quickly iterate and test and apply it to different steps of the end-to-end ad buying experience. At present, the company's feedback and results from advertisers are very good, and the company will continue to invest to make the company's advertising system, recommendation and ranking engine have higher performance and provide better service to advertisers. Q6: How does the company view this investment despite years of losses at Reality Lab? A6: We 've been able to demonstrate that we can build large-scale social experiences within the constraints of competing platforms, but we think we can do better. From a product and mission perspective, we're about helping people build great connected experiences, and the next platform we're shaping for decades to come will enable that. From a business perspective, Reality Lab will eventually have a large commercial component. We are producing products under the framework of Reality Lab to make it a good business. Currently, Family Apps has a lot of potential value, but due to Apple's restrictions on application tracking transparency, Meta has many functions that it wants to build. The company believes that this matter is unfortunate for the industry, which will make it difficult for many people to participate in it, and will lose a lot of realization value. This practice will not be recognized by the Reality Lab. Q7: About the launch of the Threads, from the experience of using the Family of Apps platform to launch new products? A7: It may be too early to analyze the Threads. Companies have not been very successful in building stand-alone applications, Messenger the biggest exception is that Facebook internal functions have been stripped out. But what is certain is that companies still have a lot of work to do to really make Threads reach their potential, and that is to build experiences that must fit the market. And the company is optimistic that Threads has the potential to become the fifth great app in Family of Apps. Once the company believes that the Threads has grown to a feel-good level, the company will give enough investment to help it grow. Q8: Information on project and equipment delays in 2023? What is the significance of the increase in capital expenditures in 2024? A8: The downward revision of the forecast for 2023 is driven by cost savings, especially the company has been looking for effective ways to allocate underutilized capacity of non-AI servers to various needs. At present, the company is still working out its capital expenditure plan for 2024, which has not been fully determined, but * * it is expected that the capital expenditure in 2024 will be higher than that in 2023 * *, because the expenditure on data centers will increase in 2024 * * as the company announces a new data center architecture at the end of 2022 to increase site construction efforts * *. At the same time, the company also expects to invest in AI workload servers in 2024 to support the company's work in AI ranking, recommendation and generative AI development. Companies will also invest in non-AI workloads as they update some servers and increase capacity. Q9: How do you see the future evolution of the AI market? How do companies choose between the two forms (open source/closed source)? What about strategic decisions about open and close source? A9: The open source/closed source AI model will continue to exist, and different companies will have different choices. Business reasons: Some companies have a business model of building a model and then selling access to that model, so open source would undermine their business model. The business model of Meta is different from this. By sharing the model, Meta can improve the quality of the model, which is a victory for Meta's business. Security reasons: There are some worries that once the capabilities of AI reach a certain level, it will be dangerous to have them openly in everyone's hands, but companies don't think that's the case today. However, we still need to be vigilant about this, so the company is not saying that it will open source everything in the future. The company has left room for further understanding of the evolution of the security situation in the future-if AI does cross certain key thresholds in the future, the company will stop open source. Q10: With regard to Q3 guidance, what are the drivers supporting the steepening YoY growth expectations? A10: On the demand side, Q3 revenue fell 4.5 per cent YoY last year, when the company did experience a period of significantly weaker demand during the year. In terms of exchange rates, we also expect the currency to be a headwind of 1 point from the previous quarter. In terms of investment, the company has been investing for a long time, the core realization work is very good, participation continues to improve. The company therefore believes it will benefit from these investments. Regarding 4Q23, we have not yet shared its revenue expectations, but there are some drivers for 4Q23's YoY growth. **At current exchange rates, * *. In addition, although advertiser demand is currently strong, last year saw considerable fluctuations in advertiser demand, so this is an unstable period and forecasting will be more difficult. Q11: How will the anticipated new product impact 24-year revenue growth? Will the new product be the core of the 2024 capex? A11: , specifically the following components -1. Core AI powers the ranking and recommendation system, supporting content ranking, engagement growth, and monetization efforts. The company is very satisfied with the return on investment in this area and will continue to invest in this area in the future to drive revenue growth. 2. * * Next-generation AI research and development around generative AI * *. The company has already established training clusters and reasoning capabilities in this area, but it is not yet certain that the 2024 work plan is yet, because there is no large-scale deployment of consumer-facing business functions, and the scale of adoption of these products will ultimately determine how much capacity is needed. However, AI computing power can be flexibly allocated, and when some of the capabilities of generative AI are no longer needed, companies can allocate these computing power to core AI. Companies want to reduce capital density in the long run. Dolphin Investment Research Disclaimer and General Disclosures