AI 公司 Anthropic 提交了 S-1 上市申请文件，预计将于下月在纳斯达克开始交易 2。分析师估算其估值区间在 1.5 万亿至 2 万亿美元之间 2。尽管表面业务涉及 IT 合同销售，但评论指出 Anthropic 的财务策略旨在证明其在算力上巨额支出的合理性 1。文章将这一行为比喻为通过类似构建“机器神”的高成本研发来避免破产 1。这种展示超级智能大规模投入以说服华尔街的模式，被视为该公司生存的关键逻辑 1。
具体财务数据显示，Anthropic 展现出极高的效率与盈利能力 2。SemiAnalysis 分析师 Joey Brookhart 指出，该公司年化经常性收入（ARR）已超过 700 亿美元 2。其推理业务毛利率从两年前的负 94% 提升至现在的正 85% 2。每兆瓦推理收入达到 6500 万美元，相比去年同期实现了十几倍的增长 2。目前，Anthropic 持有的现金储备加可投资证券超过 1200 亿美元，而在上市前仅烧掉约 100 亿美元现金 2。然而，S-1 文件披露的未来 10 年算力承诺金额约为 5000 亿至 6000 亿美元 2。
在业务结构上，Claude Code 等编码业务是当前主要收入来源 2。公司计划未来拓展网络安全、法律等垂直领域 2。虽然 Anthropic 非 GAAP EBITA 利润率为 11%，但公司倾向于将更多算力投入模型训练以追求通用人工智能（AGI），这可能压低短期利润率 2。这一长期 AGI 愿景伴随著市场竞争和财务稀释风险，构成了其商业叙事的重要组成部分 2。
Artificial intelligence firm Anthropic has submitted its S-1 filing for an initial public offering, outlining a financial strategy that justifies substantial spending on computing power in pursuit of Artificial General Intelligence (AGI). The company is expected to begin trading on the Nasdaq next month with an estimated valuation between $1.5 trillion and $2 trillion 2. Critics suggest that this heavy investment in infrastructure, metaphorically described as building a "machine god," serves as a necessary maneuver to avoid bankruptcy while maintaining the appearance of a standard IT contract business 1.
According to the filing, Anthropic reported annualized recurring revenue (ARR) exceeding $70 billion 2. SemiAnalysis analyst Joey Brookhart noted that the company’s inference gross margin improved from negative 94% two years ago to positive 85%, with revenue per megawatt of compute rising sharply to $65 million 2. Despite these gains, Anthropic retains a cash reserve plus investable securities of over $120 billion, having burned only approximately $10 billion prior to listing 2. However, the S-1 document discloses future compute commitments totaling roughly $500 billion to $600 billion over the next decade, highlighting the scale of resources required to sustain its long-term vision 2.
The company’s current primary revenue stream derives from coding tools such as Claude Code, but plans are underway to expand into vertical markets including cybersecurity and legal services 2. While Anthropic reports a non-GAAP EBITA profit margin of 11%, it prioritizes allocating additional compute resources toward model training rather than maximizing short-term profitability 2. This strategic choice underscores the tension between immediate financial returns and the ambitious goal of achieving AGI, a path that some observers argue is the only viable way for Anthropic to secure its future existence 12.
评论
还没有评论，欢迎留下第一条。