Meta Platforms 计划在2026年投入1300亿至1450亿美元的资本支出用于人工智慧基础设施建设，这一庞大的开支引发了投资者的担忧。尽管大型科技公司普遍认为在AI领域宁可过度投资也不愿错失机遇，但Meta的股价仍因巨额支出预期而遭到抛售，尤其是与微软透过Azure云端业务能将AI投入迅速转化为收入相比，Meta在变现路径上显得较为薄弱。
马克·祖克柏提出了比元宇宙时代更为清晰的AI商业计划，涵盖三大核心领域：个人AI助手、商业代理以及将资料中心租赁给其他公司。Meta去年斥资143亿美元收购Scale AI的股份，并将其创办人王亚历山大招揽至旗下，领导新成立的「超级智慧实验室」团队。该团队开发的Muse Spark模型已为Meta AI助手提供支持，使聊天机器人的用户数量增长了60%。
祖克柏认为AI生成的个性化内容将创造「近乎无限的个性化内容宇宙」，虽然这一前景带有反乌托邦色彩，却具备可观的商业潜力。目前每月约有十亿人在WhatsApp、Facebook Messenger和Instagram上使用Meta AI聊天机器人，祖克柏希望用户从单纯的对话互动转向让AI代为执行任务。整体而言，尽管AI的各项收入来源能否成功仍是未知数，但祖克柏所描绘的商业路线图比元宇宙时代更具可信度，Meta的支出也更像是经过深思熟虑的投资而非盲目的赌注。
Meta Platforms is planning $130 billion to $145 billion in capital expenditure for 2026 on AI infrastructure, a level of spending that has alarmed investors. While the prevailing wisdom in Big Tech favors overspending rather than missing a transformative opportunity, Meta's shares took a hit alongside Alphabet's as both companies signaled higher-than-expected AI outlays. Unlike Microsoft, which can translate AI spending into immediate revenue through its Azure cloud business, Meta lacks such a straightforward monetization channel.
Mark Zuckerberg has outlined a three-pronged AI strategy—personal AI agents, business agents, and data center rentals—that is more coherent than anything he articulated during the Metaverse era, which burned through over $60 billion with little to show for it. Meta's newly formed Superintelligence Labs, led by Scale AI founder Alexandr Wang after a $14.3 billion acquisition, has developed models including Muse Spark, Muse Image, and Muse Video. The Muse Spark model now powers the Meta AI assistant and has contributed to a 60% increase in chatbot users.
Zuckerberg envisions AI-generated personalized content as a powerful engagement tool, noting that AI-driven recommendation engines have already produced double-digit growth in time spent on Meta's apps. With roughly one billion people using the Meta AI chatbot monthly across WhatsApp, Messenger, and Instagram, the next challenge is shifting users from casual conversation to agentic task completion. While it remains uncertain whether these AI revenue streams will materialize, Zuckerberg's commercial roadmap for AI appears far more credible and calculated than his previous virtual-reality bet.