🤖 AI Summary
Meta’s Q3 earnings call pivoted hard to AI infrastructure: executives repeatedly invoked “compute,” “capacity,” “infrastructure” and “novel” as the company raised 2025 capex guidance and signaled even bigger AI spending in 2026. The public-facing metaverse narrative was conspicuously absent — neither “metaverse” nor Horizon were mentioned — even as internal memos from CTO Andrew Bosworth assert the metaverse remains a company priority. Investors reacted nervously, sending shares down double digits; Meta also booked a $15.9 billion tax charge tied to recent legislation, which weighed on results.
Technically, the call telegraphed a bet on large-scale compute and frontier model development rather than consumer metaverse marketing. Zuckerberg said “compute” 14 times (Susan Li 8), “infrastructure” and “capacity” each came up 12 times, and he emphasized “novel” AI capabilities seven times, framing work on a superintelligence lab and frontier models with original use cases. Hardware got attention too (Quest and “glasses” mentioned), while competitors like OpenAI, Anthropic and Google went unnamed. The signal for AI/ML practitioners is clear: Meta is reallocating resources to heavy-duty training and inference infrastructure and frontier-model R&D, accepting short-term investor pain for a long-term compute-first strategy.
Loading comments...
login to comment
loading comments...
no comments yet