Meta Platforms (META) Morgan Stanley Technology, Media & Telecom Conference summary

AI research and infrastructure

• FAIR team has contributed to open-source AI for a decade, including PyTorch and Llama models.

• $60–$65 billion invested in AI infrastructure this year, including a 2 GW data center.

• Llama 3 released with open weights; over 800 million downloads, widely used by developers and businesses.

• Meta AI assistant has 700 million monthly users, integrated into major apps except Threads.

• Focus on scaling, personalization, and voice interface for Meta AI, plus Ray-Ban Meta glasses as leading AI device.

Open-source strategy and business impact

• Open-source approach attracts top talent and fosters innovation in AI safety and auditability.

• Llama models enable global businesses to create personalized content and ads, supporting hundreds of millions of businesses.

• GenAI tools help advertisers and content creators localize and personalize content at scale.

• Llama 4 aims to deliver large-model intelligence in smaller, low-latency models with multimodal capabilities.

• Meta AI is used for contextual search, ongoing personal assistance, and deep topic conversations.

Monetization and engagement

• Monetization of Meta AI is not a current priority; focus is on utility and user experience.

• AI tools already drive ad creation and business messaging, with WhatsApp business messaging at a $2B run rate.

• AI recommendation engines deliver 30% of Facebook and 50% of Instagram content, increasing engagement and sharing.

• Threads presents a new opportunity for text-based recommender systems in the West.

• Multimodal models will enhance business messaging, enabling image and audio-based interactions.