Here are the main takeaways from Meta’s Q1 2024 earnings call as well as my assessment;
-
CEO Mark Zuckerberg said he thinks they are well-positioned to navigate the macro uncertainty.
-
CFO Susan Li said there is macro uncertainty and that the $3 billion in the guidance range aims to factor it in. She added that they have started ad spend in US pullback from China-based advertisers.
“We have seen some reduced spend in the US from Asia-based e-commerce exporters, which we believe is in anticipation of the de minimis exemption going away on May 2nd. A portion of that spend has been redirected to other markets, but overall spend for those advertisers is below the levels prior to April. But our Q2 outlook reflects the trends we’re seeing so far in April, which have generally been healthy. So it’s very early and hard to know how things will play out over the quarter, and certainly harder to know that for the rest of the year.”
-
Li said even with the capacity they are bringing in 2025, they are having a hard time meeting the compute demands across the company. She added that CapEx situation is dynamic and they continue to find a lot of good use cases to put capacity towards their AI ranking and recommendations work.
-
Zuckerberg said that in building Llama 4, they focus on low latency and context window, which are important in Meta AI voice conversation and personalization efforts respectively.
-
Zuckerberg said in the last six months, improvements in the ads recommendations have led to a 7% increase in time spent on Facebook, a 6% increase on Instagram, and 35% on Threads.
Assessment of the earnings and their AI strategy
Meta’s Q2 2025 revenue guidance ranges from $42.5 billion (+9% y/y) to $45.5 billion (+16% y/y). The lower end is in line with my own conservative estimate and remains a realistic target given the current macro uncertainty. However, the fact that Meta is confident enough to guide to the top end of that range underscores the resilience they are seeing in their ad business.
Based on management’s comments, compute capacity remains scarce. With five months of the year already behind us, and given Meta’s continued emphasis on improving ad recommendations and Llama models, I would not be surprised to see 2026 become another year of heavy CapEx investment. That said, the ROI Meta is achieving—particularly through enhanced ad targeting and conversion efficiency, may justify this level of spend.
Zuckerberg’s remarks in the earning call and in his interview with Dwarkesh Patel have strengthened my confidence in the company’s AI strategy and execution. His statement that Llama models are being designed specifically for Meta’s products reduces the relevance of comparing Llama to peer models based on generalized benchmarks. Instead, the focus should shift to real-world metrics like latency and context window, which are relevant to Meta AI personalization strategy.
I was starting to doubt Meta’s credibility following reports that Llama was optimized for leaderboard performance on LmArena. However, Zuckerberg’s clarification suggests that these optimizations are intentional and aligned with how the models will ultimately be deployed in Meta AI products. While this may limit their utility for third-party developers, it does not diminish their value within Meta’s own ecosystem. The recent launch of the Llama API should help address some of the performance and optimization gaps experienced by developers in third-party ecosystems.
Importantly, third-party validation has also helped. Artificial Analysis independently replicated Meta’s published results and concluded that the Llama 4 Scout and Maverick models are indeed among the best in their respective classes.