自ChatGPT引发生成式AI浪潮三年来，“Magnificent Seven”科技巨头的市值已增加12万亿美元，其合计市值约达20万亿美元，约等于中国GDP。这一时期也出现了行业集中：许多颇具前景的创业公司被大型既有企业通过低调收购吸收，强化了市场集中度。Anthropic虽有大科技公司的资金支持，却似乎规避了这一命运，成为常见“blitzscaling”模式的反例。其首席执行官Dario Amodei被描绘为反复对那些他认为风险过高的机会说“不”，尤其在安全问题上，而非“赶快拥抱所有机会”。

Anthropic的股权结构与其竞争对手不同：Amazon据称持有其15%至20%的股份，而Alphabet持有14%。相比之下，Microsoft持有OpenAI 27%，并与其深度绑定——Azure是OpenAI的云基础设施，Microsoft的Copilot基于OpenAI模型。Anthropic的产品策略更集中且受控：Claude目前不生成图像，降低了深度伪造风险；且其重点面向商业客户而非消费者，减少了支持海量用户所需的高昂算力成本。Bloomberg报道称Anthropic预计可实现接近200亿美元的年化收入。

在财务层面，规模也增强了其议价能力。Anthropic完成30亿美元Series G融资后，后估值达到3800亿美元，这为其在与五角大楼讨论时拒绝签署不用于自主武器或监视美国人的保证提供了分量。相比之下，OpenAI最近与五角大楼合作后快速收缩，并承认协议执行不够周全。文章认为，真正的竞争不仅带来价格压力，还能打破既有企业的意识形态惯性，迫使其回答安全与伦理问题，在一定程度上给“move fast, break things”式扩张增加阻力，从而提升AI市场健康度。

In the three years since ChatGPT ignited the generative-AI boom, the “Magnificent Seven” tech giants have added $12 trillion in market value, with a combined valuation of about $20 trillion, roughly comparable to China’s GDP. The same period also showed greater concentration: many promising startups were absorbed through stealth acquisitions by incumbents, deepening industry consolidation. Anthropic, despite backing from large tech firms, has avoided this fate and is presented as a counterexample to the usual “blitzscaling” playbook. Its CEO Dario Amodei is depicted as repeatedly saying “no” to what he sees as high-risk opportunities—especially around safety—rather than rushing to seize everything.

Anthropic’s ownership profile is different: Amazon is believed to own 15% to 20%, while Alphabet has 14%. By contrast, Microsoft owns 27% of OpenAI and is deeply integrated with it—Azure serves as OpenAI’s cloud infrastructure, and Microsoft’s Copilot is built on OpenAI models. Anthropic’s product strategy is more focused and controlled: Claude currently does not generate images, reducing deepfake risk; it prioritizes enterprise customers instead of consumer scale, avoiding the high compute costs of serving massive users. Bloomberg reports that Anthropic is on track for nearly $20 billion in annualized revenue.

Financial scale has also strengthened its bargaining power. After Anthropic raised $30 billion in a Series G round, it reached a $380 billion post-money valuation, giving it weight when refusing Pentagon demands for commitments that its model not be used for autonomous weapons or U.S. surveillance. OpenAI’s recent Pentagon engagement and rapid reversal, followed by an admission that the deal was “sloppy,” are presented as a more opportunistic path. The article argues that true rivalry adds not only pricing pressure but also forces incumbents to answer questions on safety and ethics, adding friction to a “move fast, break things” logic and potentially improving AI market health.