Story Stocks^®

Meta Platforms (META) is reportedly in discussions with Google (GOOG) about potentially using its Gemini AI models to enhance ad targeting, according to The Information. Some META employees have even considered tweaking Gemini and Gemma models with META’s own ad data, which, if true, suggests internal concern that META may be lagging GOOG in the broader AI race.

• A META spokesperson quickly downplayed the report in a Threads post, saying, “No, what’s happening here is part of the work we regularly do to evaluate third-party tools for the purpose of benchmarking.”
• This response comes against the backdrop of META’s strong in-house AI momentum. In Q2, advertising revenue grew 22% yr/yr, driven by a 9% increase in average ad prices and an 11% gain in impressions across its Family of Apps.
• META’s proprietary AI innovations -- including Advantage+, Andromeda, and GEM (Global Engagement Model) -- have delivered meaningful improvements in ad targeting precision and return on ad spend, helping sustain advertiser demand.
• The company continues to double down on its own AI roadmap, with FY25 capital expenditures projected at $66–$72 bln, much of it earmarked for Meta Superintelligence Labs. The initiative, co-led by Scale AI’s Alexandr Wang, is focused on advancing personal superintelligence technologies.
• If META were truly leaning on GOOG for core ad-targeting AI, it would mark a sharp reversal from this aggressive internal investment trajectory and raise questions about confidence in its proprietary stack.

Briefing.com Analyst Insight:

While the headline of META testing Gemini models fuels speculation about competitive gaps, the company’s clarification makes sense: benchmarking rivals’ tools is standard practice. With ad growth accelerating and proprietary AI engines like Advantage+ and GEM powering returns, META has little incentive to cede ground to GOOG in its core business. Instead, the story underscores heightened scrutiny of META’s AI efforts as competition intensifies. Investors should view this episode less as evidence of dependency on GOOG and more as a reminder that the AI arms race is forcing even leaders like META to constantly stress-test their platforms.