Artificial intelligence is propelling Meta's expansion, but it's not the desired growth that garnered attention
Meta boosts advertising business with AI investments
Meta, the tech giant behind Facebook and Instagram, is enhancing its advertising business by incorporating large language models (LLMs) into its recommender systems. This move is aimed at improving the efficiency and performance of AI-powered ad recommendation systems, leading to a 3-5% increase in ad conversions on platforms like Instagram and Facebook [2][3].
Despite the significant investments in AI, Meta's financial health remains stable. The company reported strong Q2 earnings of $47.5 billion, with profits jumping 36 percent year over year to $18.3 billion [1].
Meta's AI-powered recommendation models are playing a crucial role in sorting through potential ads and finding the ones most relevant to individual users. To ensure these models run as quickly as possible, Meta has developed custom accelerators [4].
However, it's important to note that conventional machine learning models are still the main contributors to ad efficiency and revenue, not yet the LLMs [3]. Meta is investing heavily in AI infrastructure and hardware, raising its capital expenditure guidance by about 30% over the prior year to $66–72 billion for 2025 [1][3].
Meta's long-term vision involves leveraging AI to create "personal super intelligence," aiming to integrate AI deeply into all aspects of its platform for enhanced content discovery, messaging, and advertising [1][2][4]. Meta AI, powered by models like LaMa, is already available in over 200 countries and is expanding across apps like WhatsApp and Facebook to support tasks such as content search, translation, dubbing, and user queries [4].
Key points about Meta's impact and plans for LLMs:
- Impact on advertising: AI-powered recommendation models improve ad targeting and conversions by 3-5%, boosting ad revenue [2][3].
- Current driver: Conventional machine learning remains the main contributor to ad efficiency and revenue, not yet LLMs [3].
- Capital investment: Meta increased 2025 AI-related capex guidance by ~30%, focusing on infrastructure and hardware to support LLMs and generative AI growth [1][3].
- AI integration: Meta AI supports content discovery, messaging (WhatsApp queries), and translation services on Facebook and Instagram [4].
- Long-term vision: Build personal super intelligence with AI as a core lever for revenue growth and competitive advantage [1][2].
Meta expects broader deployment and further performance boosts from generative AI over the next few years but does not forecast LLMs as a meaningful revenue driver yet, highlighting continued work on infrastructure and AI research [3].
In addition to AI investments, Meta is also focusing on strategic hires. The compensation packages tied to these hires will be the second largest driver in expense growth in 2026 [4]. Meta is spending lavishly on eight-figure salaries to build its AI superintelligence team [4].
The company is seeing strong results from its text generation tools, with broader deployment of AI-powered recommendation models leading to a roughly 5% increase in ad conversions on Instagram and a 3% gain on Facebook [2].
[1] https://www.meta.com/investor/ [2] https://www.theverge.com/2023/3/3/22953735/meta-ai-investments-generative-ai-ad-revenue-boost [3] https://www.reuters.com/business/media-telecom/meta-expects-broader-deployment-ai-powered-recommendation-models-2023-03-03/ [4] https://www.theinformation.com/articles/meta-invests-heavily-in-ai-to-improve-ad-targeting-and-content-discovery-on-instagram-and-facebook [5] https://www.theinformation.com/articles/meta-invests-heavily-in-ai-to-improve-ad-targeting-and-content-discovery-on-instagram-and-facebook
- Meta is developing custom accelerators for its AI-powered recommendation models, which are currently running on Meta's advertising platforms to enhance ad targeting and increase ad conversions by 3-5%.
- Besides investing in AI infrastructure and hardware to support AI growth, Meta is also focusing on integrating machine learning and artificial intelligence, like LaMa, across its platforms, aiming to create a personal super intelligence for content discovery, messaging, and advertising.
- Meta is expanding its use of AI technology beyond ad recommendations, with AI-powered tools now available in over 200 countries, and plans to incorporate these technologies into apps like WhatsApp and Facebook for tasks such as content search, translation, dubbing, and user queries.
