Meta Earnings Call: Strong Ad Growth, AI Integration Expands

Meta reported Q2 revenue of $47.52 billion, up 22% year-over-year, beating analysts’ expectations of $44.83 billion.

Earnings per share (EPS) came in at $7.14, a 38% year-over-year increase, well above the consensus estimate of $5.92.

For guidance, Meta expects Q3 revenue between $47.5 and $50.5 billion, ahead of Wall Street’s estimate of $46.14 billion. Management noted that Q4 year-over-year growth will likely slow compared to Q3, as the 2024 U.S. presidential election created a high ad spend base last year.

On capital expenditures, Meta raised the lower bound of its 2025 Capex outlook to $66–72 billion, up from a previous lower bound of $64 billion. The upper bound remains unchanged. Capex in 2026 may exceed $100 billion, and Meta may seek external investment to support large-scale data center expansion.

In its core AI operations, Meta continues to see strong returns on investment.

The company anticipates that both headcount and compensation will rise in 2026, as it aggressively hires and retains top AI talent with competitive pay.

Following the earnings release, Meta shares jumped 12% at the open.

📊 Social Advertising: AI Drives Better Conversion, Pricing, and Engagement

Meta’s social dominance remains unmatched, with over 3.4 billion people using at least one Meta app daily.

AI-powered ad recommendation models drove approximately 5% conversion growth on Instagram and 3% on Facebook. Average ad pricing on Facebook rose 9%, mainly due to improved ad effectiveness.

AI significantly enhanced content recommendation algorithms, helping better capture user interests. User time spent on Facebook rose 5%, and Instagram rose 6% this quarter.

Video engagement on both platforms also surged. Instagram watch time grew over 20% year-over-year, and Facebook video watch time in the U.S. also rose over 20%.

A notable portion of ad revenue now comes from generative AI tools, which are especially valuable to small advertisers with limited budgets. Nearly 2 million advertisers now use Meta's generative AI ad tools.

The company is also testing AI-powered translation, allowing ad headlines to be translated into 10 languages to support global expansion. Though generative AI is not expected to drive major near-term revenue, Meta remains very optimistic about its long-term profit potential.

On broader monetization, WhatsApp's paid messaging and Meta Verified subscription continue contributing. However, WhatsApp's ad pricing remains lower than that of Facebook or Instagram, partly due to its focus on lower-monetization regions and limited targeting data.

Meta has now begun offering video and image ads on Threads. Although Threads monetization is still in early stages, the company remains optimistic about its long-term potential as community activity increases.

🤖 AI & Large Models: Over 1 Billion MAUs, Selective Open Sourcing

Meta’s in-house AI assistant now has over 1 billion monthly active users. As Meta continues to refine its models, user engagement continues to rise.

Internal teams are experimenting with building autonomous AI agents using Llama 4, aimed at improving Facebook's recommendation algorithm, user engagement, and monetization.

Zuckerberg emphasized that while Meta is committed to open-sourcing select models, it won’t open-source everything, citing the need to stay competitive. Still, the goal remains to develop and share industry-leading open-source models in the future.

🕶️ VR & Smart Glasses: Strong Demand, Potential Everyday Essential

Meta’s Reality Labs division reported accelerating growth in Ray-Ban Meta smart glasses, with the best-selling models consistently out of stock. While Quest headset sales declined, existing users are spending more time in the Quest ecosystem, indicating strong user stickiness.

Zuckerberg noted that the smart glasses are stylish and offer fun, engaging features, making users willing to wear them all day. Usage of Meta AI on the glasses continues to rise.

He went further to say that in the future, people without smart glasses—or another means of AI interaction—might be comparable to being illiterate today. He believes smart glasses will become a basic AI interface device, perhaps one day as essential as smartphones.