I typically don’t comment on Meta’s earnings reports because I work there.
However there were several key ideas of note about AI.
Mark said Meta expects >10x improvement in Muse’s compute cost efficiency next year. Susan cited smaller models, higher GPU utilization and inference optimizations.
10M+ people set up Muse in the US and Canada in ~6 weeks, with about 25% using it on any given day.
Shopping is its most popular use case, with over 25% of weekly active users engaging in shopping-related tasks.
To me, this seems like clear evidence of consumer demand for AI agents beyond chatbots.
Susan Li made an interesting point about enterprise AI.
Some of today’s leading companies didn’t exist a few years ago.
If startups can establish themselves so quickly, why can’t Meta?
It has competitive models, infrastructure and an estimated 500M small businesses using its apps.
All of the above is taken from yesterday’s earnings call.
https://earningscalls.dev/transcripts/meta-platforms-inc_meta_earnings_call_transcript_2026-10-21



















