Skip to content
For Meta, AI Is Working at Scale; Superintelligence Would Supercharge That Growth
Meta
Meta's blowout quarter was the best example to date of AI having a tangible impact on revenue and earnings growth at scale. The question of whether AI has an ROI can be put to rest. Going forward, Zuckerberg is staking a claim to superintelligence, a dynamic in AI that experts debate in terms of feasibility. If he nails superintelligence, Meta shifts from an ad company to an AI powerhouse.

Key Takeaways

Meta reported a 22% y/y revenue increase in the June quarter, above consensus expectations of roughly 14% growth. This marked an acceleration from the 16% growth in March and reflects broad-based performance across the business. For September, they guided revenue up 21% vs. the Street at 14%.
Zuckerberg’s discussion of superintelligence on the earnings call signals to investors that the company’s ambitions in AI go beyond improving ad targeting. They want to play the long game to be a leader in general AI.
Meta is the best example of AI having a measurable impact on revenue and earnings growth.
Capex outlook for next year blew through expectations, signaling the company is serious about building out AI for the long haul.
1

Revenue

Advertising revenue rose 21% (22% in constant currency), with outperformance in key regions like Europe and the Rest of World (up over 20% y/y). Management noted that demand from advertisers, especially in the online commerce sector, including China-based companies, drove much of the upside, aided by improvements in ad targeting and measurement through AI. Pricing improved, up 9% y/y.

2

Superintelligence

Zuckerberg is all in on superintelligence, saying we all need to take it seriously. That message spoke to the reality that most industry watchers believe superintelligence, the ability of a machine to reason above a human, is not achievable.

On the call, he outlined a long-term vision for Meta in AI, saying its systems that surpass human intelligence across the board are “now in sight.” He argued that superintelligent AI could bring economic and scientific advances, as well as empower people to be more creative and connected.

Investors liked what they heard on the topic, with shares moving 2% higher when he mentioned we need to take superintelligence seriously.

3

AI on Engagement and Ad Targeting

A theme of the quarter was Meta’s use of AI to boost user engagement and advertising efficiency. Management attributed the June quarter’s performance in part to AI-driven improvements in content recommendations and ad delivery.

On the user side, Meta is doing a better job of surfacing content people find relevant. Zuckerberg explained that advancements in recommendation AI led to a 5% increase in time spent on Facebook and 6% on Instagram in the quarter. People are spending more time on Meta’s platforms because AI is showing them more relevant material, a factor behind the engagement growth mentioned earlier.

On the advertising side, AI is translating into better outcomes for marketers. During the quarter, the company expanded a new AI ads recommendation model that resulted in roughly 5% more ad conversions on Instagram and 3% more on Facebook. This means Meta’s ads are becoming more effective at prompting user actions.

The company is also using AI to automatically optimize ads, with a small portion of Meta’s ad revenue now coming from campaigns using generative AI creative tools (Advantage+ AI suite) that create or modify ad copy and imagery.

As for expenses, they grew 12% y/y (compared to revenue up 22%), a tangible example of AI reducing expenses. One reason Zuckerberg pointed out is the machine increasingly improving the machine, an observation we have not heard from the other mega-cap tech companies.

4

Investing in AI Infrastructure and Talent

Meta is ramping up capital expenditures to support AI development. CFO Susan Li suggested that Meta plans to increase CapEx again in 2026 by about 48% y/y. Prior expectations were for a 10% rise in 2026 CapEx.

They view AI hardware as a key asset: having “sufficient compute capacity will be central to realizing many of the opportunities in front of us,” Li noted. In practical terms, Meta is building some of the world’s largest data centers and AI clusters. Zuckerberg mentioned that the Prometheus AI training cluster coming online in 2024 will likely be the first to exceed one gigawatt, and a follow-on system called Hyperion is planned to scale up to 5 GW over time. The arms race is on, especially when compared to expectations that xAI has for Colossus 2.

As for investing in AI talent, it has been well documented that in the past two months the company has been hiring from the industry’s top ranks. Li noted that while headcount growth in most other areas of the company remains constrained, Meta is prioritizing AI hiring even if it drives up expenses. The compensation for these AI hires is already built into Meta’s expense outlook for 2025 and 2026, which calls for the growth in expenses to accelerate next year, an update that investors read positively, with shares moving 1% higher after hours on the news.

Disclaimer

Back To Top