Meta Platforms, Inc. (META) Earnings Call Transcript & Summary

Dear guests and business leaders, it's an immense pleasure to be here in this room, such a big room, exactly 1 year after the previous Deloitte Enders conference last year. A lot happened in this year and a lot is happening in the world. When I talk to business leaders like yourselves, a few things are top of mind really. The radical uncertainty that is going on in the world with all the geopolitical tensions, trade and tariff wars, the stagnation that we see in the global economy, we are probably going through the slowest half decade over the last 30 years. And all the demographic changes that is happening, aging populations, today, in Europe, 1/5 of the population is above 65 years old. This is expected to double in 2060. So those people will be going out of the workforce. And this is coming at a time that governments are already struggling, providing good quality public and health services to aging populations while struggling with fiscal deficits that started to accumulate in 2008 crisis and then exacerbated by COVID as well. But it's not all doomsday scenario. There's also a lot of technological advancements in the world. And we do believe that AI will have the power to address many of these societal and economic changes that is happening in the world right now. We believe that it will transform fundamentally the way that people think, decide, problem solve and it's going to enhance human creativity. I mean we talk about AI every day, and these are all the things that we know, but it is going to advance technological and scientific breakthrough and enable access to information and education in every single language. We are working on models that translates 1,100 languages like unheard languages real time. And all of these things is going to increase the productivity and add trillions of dollars into the global economy. Goldman Sachs estimates that it's going to increase the global GDP by 10% over the next 10 years, which is a very big number. So I think it's safe to say that our lives is going to be fundamentally changed in the way that we work, we learn and interact with one another. And obviously, in all of that, all companies play a role and we too. And I would like to talk to you about a little bit about how we see this future at Meta as well. Talk about our North Star vision. Actually, we don't have one vision, but actually two. So it's building the AGI, the Artificial General Intelligence and building the Metaverse as the next computing platform that people will interact. And we talk a lot about AGI. I'm sure everybody has heard about the word, and there's a lot of debate on how do we define it, how will we know when we reach AGI and a lot of conversations around this. And I would like to double-click on that a little bit. The way that we define AGI is basically AI reaching human-level capabilities. So machines being able to reason, draw conclusions and to take actions at a human level or beyond in a complex set of tasks. So in order to achieve that, there will be 3 main pillars that need to come together in order to get us there. The first pillar is going to be the foundation models, and I will talk about each of these pillars. The second pillar is how we are evolving from assistants to agents and to be able to build these layers on top of the foundation models. And the third layer is the wearables, all the AR/VR devices and how this incredible technology is really being embedded into our day-to-day lives through the devices and robotics as well. So let's start with the foundation models. So the -- in order to get to AGI, we are working on 6 different buckets, let's say, building blocks in building the foundation models. The first one is the perception. So this is basically how the models understand and interpret the data from their environment. The other one is the communication. So how do models express themselves and how do they understand what others are talking about through using language. The third one is reasoning and how models problem solve, draw conclusions, make decisions and take action. And the alignment work is bringing models closer to human and societal values. And the embodiment, which is exciting, is how we are bringing this technology into wearables and from there, moving into robotics as well. And all of that, all the foundational layer is around the systems, which is the infrastructure and the data that we need to move all of that. So in order to reach AGI, the work needs to happen in these 6 building blocks simultaneously to move us there. And when we talk about the models, our foster child is Llama. And this is pretty much all the platforms that you use from Meta; Instagram, Facebook, WhatsApp, Meta AI is all built on Llama. And just to talk about the speed of the technology, Llama was a research project in 2022. We launched Llama in February 2023. In 2 years, today, there has been 1.2 billion downloads of Llama. It's the widest used open source model globally. And I presented this in Paris AI Action Summit in February. This number was 700 million. So that's the accelerating speed on how these models are being made available to developers, to the entire ecosystem in order to advance and enrich this ecosystem together. And this next generation of models is going to be bringing new capabilities, particularly around multimodality, which means that models are not able to understand text or language-based input, but it's also image and sound and movements in the environment that they operate, hope -- seeing that through wearables as well, and also enhanced image reasoning and being able to build lighter versions so they can be customized for different use cases and be used on local devices as well. And there is already a vibrant community of developers around it, and there has been 200,000 derivatives on Hugging Face. So this is basically developers building their own derivatives using our models, and that ecosystem is enriching through the partnerships that we are doing. The second layer of this is the layer of assistants and agents. I'm sure everybody has heard about the agentic world and -- but what is the difference really? We are going through a massive foundational shift from word models to world models. And for non-native speakers, it's very difficult for me to pronounce this. So word models are based on language, based on text, and they can understand what we write. And even if they are imagining a glass, they understand because it's described through text, right? But this is not how human beings learn. When a baby comes to the world, like they don't start speaking. They start observing. They are learning through visual input. And how we are going to move toward models is through the wearables and being able to help AI understand the environment that we operate. And that is going to make the change to -- from assistants to agents. Today, we are talking about assistants, Meta AI, for example, our Meta's assistant and every big tech company has their own assistant and their customized assistants. And what they do is they understand the questions you ask. There are particular use cases. They can help you prepare a presentation, and I used a little bit for this presentation as well. And they can help you interact with the people around you. But what is the difference to agents? So that is the word models moving into world models. So agents are not only able to understand and respond to what you say, they are able to take action on your behalf. And they are able to work with other agents in the background and to do that as well. And this is exciting. We believe that at some point, every single business will have their own agent. And many companies that I talk to today, they are already working on their agentic strategies. And at Meta, we plan to do this as seamless and simple as we did in the ads business as well in order to make it accessible to all businesses from small to big. But the use cases for the agents are not only for businesses. It's also in our day-to-day lives. And just think about this. My agent is connected to my wearable that's tracking my health, and it's waking me up, looking at my sleep metrics, waking me up exactly at the time that I feel most refreshed. And when I'm having my healthy and beauty sleep, my personal agent is going through the school's newsletters and the 5 different apps that they are using for homework, for community engagement, for all of those things and already went into my e-mail and to Amazon to put everything in the basket to -- when I wake up, it's asking for my approval. These are the things that kids need to take to school in this last week of school. There are many things. And I can just say, okay, and it's purchased from Amazon and sent to the school a week before the deadline. While my personal assistant is doing that, my work agent -- sorry, personal agent is doing that. My work agent has already gone through all the e-mails overnight and my work chats and prioritize them based on importance and urgency and already proposed a particular way that I should spend my time working on which tasks and how much time. So I'm very efficient and high impact so I can spend time with my kids at dinner time. So for a time-poor working mom like me, that sounds amazing. And we do believe that with the right foundations, we will be able to quickly move this from digital to wearables and robotics as well, which takes me to my next pillar, which is the wearables. And we do believe that people in the future will want to engage with one another, not through phones, but in a different form factor, a form factor that allows us to engage with our AI, our personal agents, our work agents seamlessly without having to take the phone from our pocket and write something like the ones that I'm wearing today. And these glasses, they are already available in the market and can do stuff like this. Hey, Meta start. By the way, I shouldn't say it because it's going to activate. Hey, start my running playlist on Spotify or how can I pair this wine or who is this person on the statute. So it's already understanding visual input and can help me before -- without the need of friction of using a phone and trying to put my reading glasses and trying to understand what I'm doing, right? So the next step of where these glasses are going to evolve is what we launched last year in Connect in September, which is a prototype so that developers can work with. It's called Orion glasses. And if you allow me, I have a 1-minute video to show you how they look like and how we build this been in the making for nearly 10 years. And yes, it's a glimpse into the future. So I'll ask AV team to run the video, please. [Presentation]

So that's how it's going to look like. We showed the demo in Davos. And if anybody is going to Cannes, we will also have demos there as well. So you're wearing this wrist band, which can read your electrical waves in your wrist. And with very subtle moments, you can just basically see screens within this glass and open the fridge and ask what can I cook with the ingredients that I have, and it's just going to give you the recipes already and instructions to cook, et cetera. So that's kind of the future we are moving towards. But more than this cool technology, what makes me most proud is the open source approach we have to building this. And we are investing approximately $70 billion into this technology annually and with the heart -- open source at the heart, making this technology accessible to all people, all developers, all businesses, big and small as well. So we believe that open source is going to lead to better innovation, safer innovation, better economic growth because broader accessibility to this technology is going to accelerate innovation and economic growth, while the quick discovery and detection of risks, not only post training, even during training, having more eyes is going to lead to safer technology and more responsible development as well. But just thinking about the future, it's really important to make sure that we remember the economic value and the productivity that these technologies are creating in the world. We published an economic study a couple of weeks ago. In Europe, specifically, there is $213 billion of economic activity linked to our platforms just on the advertising side. I'm not talking about the assistants and the open models that we deliver. That means 1.4 million jobs and 99% of those jobs are small businesses. So this is the type of impact we are talking about. And with these opportunities, though, comes barriers, limitations and challenges as well. And if we are not careful, the very restrictive regulations, I'm not saying regulations, the restrictive ones, particularly in Europe, which is very fragmented, inconsistent and rigid, might also hamper this innovation and all the technologies that I've shared today. It's not only about Meta. I mean we brought the Meta AI to Europe 18 months after we launched it in Nigeria. And it's not only us, it's Google's latest technologies and Apple Intelligence, Microsoft's Copilot, they all launch late. So it means that we are not empowering European businesses with the latest technologies. So it will be really important to advocate for balanced policies that is going to support innovation, but also encourage responsible build development as well because the choices that we make today about data, technology and policy will determine whether the Internet for tomorrow is going to be open, innovative and valuable for everyone. Thank you.