Mark Shmulik

Zuckerberg engaged the premise directly, contrasting the technical coding-agent market with the simplicity bar for billions of consumers.

Zuckerberg clearly described the two parallel loops: research team scaling models and product team building on top of them. Both iterate continuously.

Li declined to give 2027 CapEx guidance but acknowledged they have continued to underestimate compute needs and expect compute to become even more central.

Susan Li explicitly clarified the commitment is absolute OI dollars above 2025, not OI growth rate, an important distinction given the expense acceleration.

Susan Li emphasized that basis-point performance gains off a large base translate into meaningful absolute revenue growth.

Susan Li signaled non-linear profit growth, explicitly warning investors that heavy-investment years will compress operating profit growth.

Zuckerberg listed qualitative internal markers but offered no external KPI, also warned of lag in monetization.

Zuckerberg reaffirmed the mid-level engineer AI milestone timeline as on track, with scaling into 2026.

Susan Li explained lowered expense reflects updated comp forecasts, offset by higher infra/RL COGS; $5B range maintained.

Susan Li reframed price-per-ad as an output metric and pointed to ongoing CPA improvements lifting CPMs over time.

Zuckerberg said AI now drives at least as much value in glasses as holographic AR, flagging 2025 as a pivotal year.

Sundar: allocation starts with frontier AGI development, then core products and Cloud serving; third-party bridge deals are ROI-positive over the multiyear horizon.

Schindler described the expansionary moment for search broadly rather than quantifying the specific contribution of AI-driven conversion improvements.

Ashkenazi described highly rigorous framework evaluating all investment needs estimating near and long term yields. Determines total investment envelope and allocation across organization. AI investments already delivering results across business with cloud most visible externally. Also considers cash flow generation balance sheet health and fiscal responsibility. Pichai said top concern is capacity constraints including power land supply chain and how to ramp to meet extraordinary demand while driving efficiencies.

Pichai said AI Overviews engagement very high as natural part of Google experience. AI Mode has varied cohorts with core group of passionate early adopters who are seeking it out. Across all use cases seeing early adopters then more people coming in with users who use it continuing to use more over time with high satisfaction. Schindler said AI Overviews monetizing at approximately same rate as great baseline. Testing ads in AI Mode with opportunity on historically under-monetized queries through smart AI integration.

Pichai said top AI people look for combination of frontier mission state-of-the-art work access to compute and working with best peers. Both retention and new talent metrics healthy despite individual headline departures. Ashkenazi said having full stack including research is critical resource. Investing appropriately in best minds while driving efficiency across business using AI tools internally. Continuing same discipline execution and efficiency efforts previously discussed.

Pichai said over 30% of code checked in now involves AI-suggested solutions up from 25%. Working on early agentic coding workflows with deeper experiences. Customer service teams leading internal adoption with enhanced user experience and efficiency. AI deployed across all company functions including finance and earnings preparation. On AI Mode people typing 2x longer queries with complex nuanced questions and product comparisons working well.

Pichai said Gemini 2.0 built for enabling agentic use cases with definite progress. Agents expand opportunity beyond information to acting on needs. Far from zero-sum game with plenty of room for new use cases. Clear sense of additional use cases for Search users. AI overview response shows users react positively to new capabilities.

Doug laid out the FCF-per-share framework: operating performance, disciplined capital allocation and dilution management; direct.

Evan connected direct-revenue traction to confidence in Specs and future monetization; direct.

Evan framed Snap's chat interface as a distribution channel for AI agents via Sponsored Snaps; direct.

Evan pointed to the ~$700M run rate and Lens+/creator products as increasingly meaningful revenue; direct.

Evan said Snap differentiates My AI through visual communication and AR as the eventual AI interface; direct.

Evan said platform uncertainty is benefiting Snap as advertisers and creators diversify; direct. (Asked by associate Luiza Nobre on behalf of Mark Shmulik.)

Jen said ad load is low versus peers and not a strategic lever; growth comes from users and more valuable impressions.

Steve said Reddit is mid-way through a top-to-bottom upgrade of people, process and tech, already seeing some retention gains.

Steve said Reddit expects a balanced mix, with most users being both scrollers and seekers in a single day.

Julia detailed UCAN reacceleration drivers and said Europe/RoW moderation was expected on tougher comps and go-to-market changes.

Julia cited the lower-funnel playbook translating abroad; Bill said the whole platform is becoming search-like.