Meta's $25B privacy penalty reprices the data asset its AI capex demands

Meta's October 21 earnings day combined a $25 billion civil penalty in New Mexico for deceiving users about data handling with a raised 2026 capital-expenditure floor of $135 billion for AI infrastructure. The New Mexico verdict, rooted in the Cambridge Analytica scandal, found 43 million violations of state consumer protection laws and initially yielded a theoretical $219.5 billion before being reduced to $25 billion. Meta plans to appeal, and the penalty applies to deceptive statements about historical data handling and hate-speech moderation, not AI training specifically. Still, the verdict demonstrates that courts can retroactively reprice how user data was collected, raising the question of whether AI training pipelines could face similar liability. Meta's core business remained strong, with Q3 revenue up 26% to $64.6 billion and ad prices up 14%, though operating income fell 37% due to a $9.89 billion settlement charge. To fund AI investments, Meta announced a new Class C nonvoting stock and raised its 2026 capex range to $135-145 billion. The market reacted positively, but the long-term risk of data-liability claims targeting AI training remains an open question.

(Source：Vector Wire)