Meta 三季度营收同比增长 26%,Muse 智能体成本高企尚未盈利
Meta Grows Revenue 26% Despite Massive Muse Operating Costs
- 1Meta Q3 应用家族营收同比增长 26% 至 642 亿美元,其中广告收入 631 亿美元,广告展示量增长 10%。
- 2Meta 背负多起诉讼,儿童伤害多州案达成 180 亿美元和解,新墨西哥州隐私案被判罚 250 亿美元,公司正上诉。
- 3Muse 智能体上线后美加超 1000 万人设置个人智能体,约 25% 每天使用,但运行算力成本高,官方称尚未盈利。
- 4官方称 Generative Recommender 上线后定向人群转化率提升近 2%,GEM 模型带来站外转化增 1.7%、应用内增 9.6%。
Meta 的 AI 变现走两条线,广告推荐模型已在贡献可量化的转化提升,消费者助手 Muse 用户多但仍是不盈利的成本中心。
- 影响:广告主需注意 Meta 平台的转化提升越来越依赖其自有 AI 推荐模型,第三方广告技术与人工创意的空间受挤压;消费级智能体算力成本高企,说明此类产品距规模化盈利尚远;巨额隐私判罚提示依赖用户数据的投放面临合规成本上升。
- 参考:投放 Meta 广告的企业可考虑把官方公布的转化提升幅度作为对照基准,分别测算 AI 优化工具与人工投放的效果差异再分配预算;评估智能体类产品时值得先核算单用户推理成本和达到盈亏平衡所需的收入规模。
- 依据:营收与 Muse 用户数据来自 Meta 三季度财报电话会官方口径,暂无独立验证;转化提升数字为 CFO Susan Li 在电话会公布的官方数据;180 亿美元和解与 250 亿美元罚单为法院裁决,Meta 正上诉。
Wednesday, October 21st, 2026 – 8:14 pm
SHARE:
In some regards, it’s been a rough year for Meta.
It’s been hit with a number of major lawsuits, including a multi-state action over the harmful effects of social media on children, resulting in an $18 billion settlement. Another suit regarding misleading users on data privacy resulted in the judge ordering Meta to pay the state of New Mexico $25 billion in penalties. The latter order came mere hours before Meta’s third quarter earnings call on Wednesday (Meta is appealing the decision).
Nevertheless, Meta had a strong Q3, all things considered.
Revenue for its family of apps was up 26% year-over-year to $64.2 billion, $63.1 billion of which came from ad revenue. Ad impressions were also up by 10%.
And Meta provided plenty of updates on how its Muse personal AI assistant ties into its ever-growing advertising business.
Something to Muse on
Meta’s earnings calls have been dominated by its AI ambitions for at least the past two years. But, at its heart, Meta is an advertising company, even more than it’s an AI company.
So, naturally, Muse has an advertising arm.
A few weeks after Muse’s initial launch in early September, Meta unveiled Muse for Small Business. It’s “a new set of new skills and connectors” in the Muse app that helps businesses find new customers, analyze ad performance, optimize ads and draft campaigns, Meta CFO Susan Li told investors.
On the consumer side, the Muse launch was one of the company’s “most successful launches ever,” said Meta CEO Mark Zuckerberg – a debatable sentiment, considering the host of privacy concerns around early versions of the model.
But much like Meta’s other platforms, privacy concerns haven’t hindered adoption.
Since Muse’s launch, over 10 million people in the US and Canada have set up a personal agent, and about 25% of those people use it on any given day.
However, all those agents are “very compute intensive to run,” said Zuckerberg. Meta expects its cost efficiency to improve by 10x over the next year, which “which will enable it to be profitable once the revenue lines are at scale,” according to Zuckerberg.
Which is to say: For now, Muse might be popular, but it’s not making money.
AI-s on the prize
So, if Meta’s increase in ad revenue can’t be attributed to Muse, then what’s responsible?
The answer is still AI, which has been central to Meta’s overarching ad platform strategy for a while now.
Earlier this year, Meta launched its Generative Recommender, an LLM-powered tool that analyzes ad content and user preferences to determine the best ad to serve a given person. Targeted audience segments have seen a nearly 2% increase in conversions since the tool was launched on Facebook feeds, said Meta CFO Susan Li.
Meta also expanded GEM, its generative ad recommendation model, to unify recommendation data across different platforms in one place. This capability resulted in a 1.7% increase in conversions for offsite campaigns and a 9.6% increase in in-app conversions this past quarter.
AI investments have been benefitting Meta’s ads business for years, said Li. As LLMs become the foundation of more AI infrastructure, she added, Meta will have more insight into “people’s preferences [and] what they’re interested in, both across organic content and recommendations.”
The goal is to not only fuel Meta’s Advantage+ ad optimization engine, but also Meta’s creative personalization features. Third-party ad tech and human creatives beware.
来源:AdExchanger · adexchanger.com