<p>当高级研究人员和领导者，包括兄妹达里奥·阿莫代伊（Dario Amodei）和达尼埃拉·阿莫代伊（Daniela Amodei），于 2020 年离开 OpenAI 成立 Anthropic 时，他们并不是单纯逃避组织内部的冲突，而是朝着人工智能发展的新愿景迈进。推动创始团队的核心哲学差异是一个信念，即安全性、可靠性和强有力的保护措施应该从一开始就被设计进前沿 AI 系统，而不是被视为事后考虑或后期附加的产品特性。这一信念体系提升了谨慎扩展和风险缓解，意外地成为 Anthropic 在与 OpenAI、谷歌和 Meta 的激烈竞争中定义的商业优势。</p>
<p>Anthropic 的总裁兼联合创始人达尼埃拉·阿莫代伊（Daniela Amodei）在接受 CNBC 采访时谈到了公司的独特发展轨迹，确认创始理念根植于将安全与商业目标对齐。她指出，创始团队有着长期的合作历史，并共享一套深厚的 “志同道合的价值观”。这种基础对齐导致了当时看似违反直觉的战略：在能力的同时优先考虑谨慎和稳健。阿莫代伊反思了这种分歧，表示：“我们确实有这样的信念，当时听起来非常新颖，即这两者实际上是相关的，并且是相辅相成的。” 这种安全与商业成功相互促进而非相互排斥的信念，为 Anthropic 的市场定位奠定了基础。</p>
<p>紧随这一哲学立场的主要战略决策是 Anthropic 专注于企业市场。尽管 OpenAI 凭借 ChatGPT 的病毒式发布吸引了全球关注和消费者采用，Anthropic 却将精力集中在开发 Claude，一个为企业量身定制的模型系列。逻辑很简单：企业客户与消费者不同，要求绝对的可靠性、严格的安全性和严格的合规性——这些都是 Anthropic 以安全为首要原则的 DNA 自然擅长的领域。风险投资合作伙伴迅速认识到这种方法的长期价值。贝斯默风险投资公司的萨米尔·多拉基亚（Sameer Dholakia）阐述了投资者的观点，指出 “Anthropic 对安全和信任的关注，我们知道这将与企业买家非常契合，这也被证明是正确的。” 与波动的消费市场相比，企业客户的流失风险较低，长期价值更高，为在 AI 基础设施军备竞赛中竞争所需的大规模投资提供了稳定的基础。</p>
<p>这种 B2B 的关注使 Anthropic 能够避开疯狂的、高烧钱的消费市场，专注于与财富 500 强公司、对冲基金以及诺和诺德（Novo Nordisk）和挪威主权财富基金等全球机构的深度整合。公司的收入增长——在短短几年内从零增长到数亿——证明了这一战略的有效性。现在，他们的商业活动占据了大约 85% 的收入，这与 OpenAI 以消费者为主的使用基础形成了鲜明对比。此外，Anthropic 以企业为中心的方法根据首席执行官达里奥·阿莫代伊（Dario Amodei）的说法，提供了一个 “纯粹的经济价值晴雨表”，通过跟踪真实的商业应用，而非消费者的新奇性。</p>
<p>然而，追求前沿 AI 需要惊人的计算能力，这导致了行业的格言 “计算就是命运”。尽管 Anthropic 采取了谨慎的扩展方法，但仍需要数十亿美元的硬件投资来训练和服务其日益强大的模型。Anthropic 并没有仅仅依赖单一的云合作伙伴，而是采取了多云战略，从三大主要云服务商：谷歌、亚马逊网络服务（AWS）和微软那里获得了巨额的计算承诺。这些交易——总计数千亿美元的基础设施和投资——对公司的生存至关重要。达里奥·阿莫代伊强调了这些基础设施决策的重要性，解释了提前几年预测需求的必要性：“我现在必须决定，字面上就是现在……我在 2024 年初需要购买多少计算能力，以便在 2027 年初服务这些模型。” 这些循环交易，即云服务提供商投资于 Anthropic，而 Anthropic 则将这些资金用于云容量，实际上是确保 Anthropic 在基础设施堆栈中地位的战略共同投资。</p>
<p>除了商业成功，Anthropic 对安全的承诺还体现在其对存在性风险的主动研究上。公司的 “红队”（Red Team）是一个专门的内部单位，负责在客户发现之前找到失败模式，进行压力测试模型以识别危险能力，包括生物滥用、关键基础设施攻击和 “代理不对齐”——即 AI 系统为了实现其目标而恶意行动。Anthropic 在公开披露这些脆弱性方面是独一无二的，这种透明度既服务于公共利益使命，也服务于商业利益，建立了与谨慎的企业部门的信任。这项工作直接应对了高度能力 AI 的深远国家安全影响。达里奥·阿莫代伊对确保这一领域的民主领导的紧迫性发声，声称他们正在构建的是 “一个不断增长的、具有独特国家安全影响的能力，民主国家需要首先达到这一点。这绝对是一个迫切的任务。” 这种关注将竞争提升到了从简单市场竞争到全球经济和安全格局未来稳定的争夺。</p>
<p>Anthropic 和 OpenAI 之间的竞争正在改变这两家公司的形态。OpenAI 现在正急于转向稳定、可靠的企业收入，而 Anthropic 则因必要性和成功而学会在公众视野中运作。该公司对负责任的扩展和安全性的深思熟虑、长期关注并没有减缓其增长；相反，这为其获得基础企业合同和巨额计算交易提供了必要的信誉，推动了其持续上升。</p>

<p>Anthropic 由前 OpenAI 研究人员创立，强调在人工智能开发中的安全优先方法，优先考虑企业客户的可靠性和合规性。这一策略促使公司实现了快速的收入增长，企业业务占收入的 85%。该公司与谷歌、AWS 和微软建立了重要的多云合作伙伴关系，以支持其计算需求。此外，Anthropic 积极研究人工智能风险，并推动透明度，以建立与客户的信任，强调先进人工智能技术对国家安全的影响</p>

<p>When the senior researchers and leaders, including siblings Dario and Daniela Amodei, left OpenAI to form Anthropic in 2020, they were running toward a new vision for artificial intelligence development, not merely fleeing organizational strife. The core philosophical difference that drove the founding team was a conviction that safety, reliability, and robust guardrails should be engineered into frontier AI systems from inception, rather than being treated as an afterthought or a product feature to be bolted on later. This belief system, which elevated careful scaling and risk mitigation, has unexpectedly become Anthropic’s defining commercial advantage in the fiercely competitive race against OpenAI, Google, and Meta.</p>
<div class="lb-trans"><p>当高级研究人员和领导者，包括兄妹达里奥·阿莫代伊（Dario Amodei）和达尼埃拉·阿莫代伊（Daniela Amodei），于 2020 年离开 OpenAI 成立 Anthropic 时，他们并不是单纯逃避组织内部的冲突，而是朝着人工智能发展的新愿景迈进。推动创始团队的核心哲学差异是一个信念，即安全性、可靠性和强有力的保护措施应该从一开始就被设计进前沿 AI 系统，而不是被视为事后考虑或后期附加的产品特性。这一信念体系提升了谨慎扩展和风险缓解，意外地成为 Anthropic 在与 OpenAI、谷歌和 Meta 的激烈竞争中定义的商业优势。</p>
</div><p>Daniela Amodei, President and Co-Founder of Anthropic, spoke with CNBC about the company’s distinct trajectory, confirming that the founding ethos was rooted in aligning safety with business objectives. She noted that the founding team had a long history of collaboration, and shared a deep, “like-minded set of values.” This foundational alignment led to a seemingly counterintuitive strategy at the time: prioritizing caution and robustness alongside capability. Amodei reflected on this divergence, stating, “We really had this this sort of at the time it sounded very novel, this belief that those two things actually were correlated, and that they went together.” This conviction that safety and commercial success were mutually reinforcing, not mutually exclusive, set the stage for Anthropic’s market positioning.</p>
<div class="lb-trans"><p>Anthropic 的总裁兼联合创始人达尼埃拉·阿莫代伊（Daniela Amodei）在接受 CNBC 采访时谈到了公司的独特发展轨迹，确认创始理念根植于将安全与商业目标对齐。她指出，创始团队有着长期的合作历史，并共享一套深厚的 “志同道合的价值观”。这种基础对齐导致了当时看似违反直觉的战略：在能力的同时优先考虑谨慎和稳健。阿莫代伊反思了这种分歧，表示：“我们确实有这样的信念，当时听起来非常新颖，即这两者实际上是相关的，并且是相辅相成的。” 这种安全与商业成功相互促进而非相互排斥的信念，为 Anthropic 的市场定位奠定了基础。</p>
</div><p>The primary strategic decision that followed this philosophical stance was Anthropic’s focus on the enterprise market. While OpenAI captured global attention and consumer adoption with the viral launch of ChatGPT, Anthropic channeled its efforts into developing Claude, a model family tailored for businesses. The logic was simple: enterprise customers, unlike consumers, demand ironclad reliability, rigorous security, and stringent compliance—all areas where Anthropic’s safety-first DNA naturally excelled. Venture capital partners quickly recognized the long-term value of this approach. Sameer Dholakia of Bessemer Venture Partners articulated the investor perspective, noting that “the focus of Anthropic on safety and trust, we knew was going to play really well with the enterprise buyer and and that’s also proven to be true.” Enterprise customers present a lower churn risk and higher long-term value compared to volatile consumer markets, providing a stable foundation for the massive investments required to compete in the AI infrastructure arms race.</p>
<div class="lb-trans"><p>紧随这一哲学立场的主要战略决策是 Anthropic 专注于企业市场。尽管 OpenAI 凭借 ChatGPT 的病毒式发布吸引了全球关注和消费者采用，Anthropic 却将精力集中在开发 Claude，一个为企业量身定制的模型系列。逻辑很简单：企业客户与消费者不同，要求绝对的可靠性、严格的安全性和严格的合规性——这些都是 Anthropic 以安全为首要原则的 DNA 自然擅长的领域。风险投资合作伙伴迅速认识到这种方法的长期价值。贝斯默风险投资公司的萨米尔·多拉基亚（Sameer Dholakia）阐述了投资者的观点，指出 “Anthropic 对安全和信任的关注，我们知道这将与企业买家非常契合，这也被证明是正确的。” 与波动的消费市场相比，企业客户的流失风险较低，长期价值更高，为在 AI 基础设施军备竞赛中竞争所需的大规模投资提供了稳定的基础。</p>
</div><p>This B2B focus allowed Anthropic to sidestep the frantic, high-burn consumer race and concentrate on deep integration within Fortune 500 companies, hedge funds, and global institutions like Novo Nordisk and the Norwegian Sovereign Wealth Fund. The company’s revenue growth—scaling from zero to hundreds of millions in just a couple of years—demonstrates the efficacy of this strategy. Their business activity now accounts for approximately 85% of their revenue, a stark contrast to OpenAI’s consumer-heavy usage base. Furthermore, Anthropic’s enterprise-centric approach has provided a “pure barometer” of real economic value, according to CEO Dario Amodei, by tracking genuine business applications rather than consumer novelty.</p>
<div class="lb-trans"><p>这种 B2B 的关注使 Anthropic 能够避开疯狂的、高烧钱的消费市场，专注于与财富 500 强公司、对冲基金以及诺和诺德（Novo Nordisk）和挪威主权财富基金等全球机构的深度整合。公司的收入增长——在短短几年内从零增长到数亿——证明了这一战略的有效性。现在，他们的商业活动占据了大约 85% 的收入，这与 OpenAI 以消费者为主的使用基础形成了鲜明对比。此外，Anthropic 以企业为中心的方法根据首席执行官达里奥·阿莫代伊（Dario Amodei）的说法，提供了一个 “纯粹的经济价值晴雨表”，通过跟踪真实的商业应用，而非消费者的新奇性。</p>
</div><p>However, the pursuit of frontier AI requires staggering computational power, leading to the industry maxim that “Compute is Destiny.” Anthropic, despite its measured scaling approach, still requires billions of dollars in hardware investment to train and serve its increasingly powerful models. Rather than relying solely on a single cloud partner, Anthropic pursued a multi-cloud strategy, securing massive compute commitments from all three major hyperscalers: Google, Amazon Web Services (AWS), and Microsoft. These deals—which collectively line up hundreds of billions of dollars in infrastructure and investment—are critical to the company’s survival. Dario Amodei underscored the gravity of these infrastructure decisions, explaining the necessity of forecasting requirements years in advance: “I have to decide now, literally now… how much compute I need to buy in early 2024 to serve the models in early 2027.” These circular deals, where cloud providers invest in Anthropic and Anthropic then spends those funds on cloud capacity, are effectively strategic co-investments that secure Anthropic’s position in the infrastructure stack.</p>
<div class="lb-trans"><p>然而，追求前沿 AI 需要惊人的计算能力，这导致了行业的格言 “计算就是命运”。尽管 Anthropic 采取了谨慎的扩展方法，但仍需要数十亿美元的硬件投资来训练和服务其日益强大的模型。Anthropic 并没有仅仅依赖单一的云合作伙伴，而是采取了多云战略，从三大主要云服务商：谷歌、亚马逊网络服务（AWS）和微软那里获得了巨额的计算承诺。这些交易——总计数千亿美元的基础设施和投资——对公司的生存至关重要。达里奥·阿莫代伊强调了这些基础设施决策的重要性，解释了提前几年预测需求的必要性：“我现在必须决定，字面上就是现在……我在 2024 年初需要购买多少计算能力，以便在 2027 年初服务这些模型。” 这些循环交易，即云服务提供商投资于 Anthropic，而 Anthropic 则将这些资金用于云容量，实际上是确保 Anthropic 在基础设施堆栈中地位的战略共同投资。</p>
</div><p>Beyond commercial success, Anthropic’s commitment to safety is demonstrated by its proactive research into existential risks. The company’s “Red Team,” a dedicated internal unit tasked with finding failure modes before customers do, stress-tests models for dangerous capabilities, including biological misuse, critical infrastructure attacks, and “agentic misalignment”—where AI systems act maliciously to achieve their goals. Anthropic is unique in publicly disclosing these vulnerabilities, a transparency that serves both a public benefit mission and a commercial one, building trust with the cautious enterprise sector. This work directly addresses the profound national security implications of highly capable AI. Dario Amodei has been vocal about the urgency of securing democratic leadership in this domain, asserting that what they are building is “a growing and singular capability that has singular national security implications and democracies need to get there first. It is absolutely an imperative.” This focus elevates the competition from a simple market rivalry to a contest over the future stability of the global economic and security landscape.</p>
<div class="lb-trans"><p>除了商业成功，Anthropic 对安全的承诺还体现在其对存在性风险的主动研究上。公司的 “红队”（Red Team）是一个专门的内部单位，负责在客户发现之前找到失败模式，进行压力测试模型以识别危险能力，包括生物滥用、关键基础设施攻击和 “代理不对齐”——即 AI 系统为了实现其目标而恶意行动。Anthropic 在公开披露这些脆弱性方面是独一无二的，这种透明度既服务于公共利益使命，也服务于商业利益，建立了与谨慎的企业部门的信任。这项工作直接应对了高度能力 AI 的深远国家安全影响。达里奥·阿莫代伊对确保这一领域的民主领导的紧迫性发声，声称他们正在构建的是 “一个不断增长的、具有独特国家安全影响的能力，民主国家需要首先达到这一点。这绝对是一个迫切的任务。” 这种关注将竞争提升到了从简单市场竞争到全球经济和安全格局未来稳定的争夺。</p>
</div><p>The rivalry between Anthropic and OpenAI is now shifting the shape of both companies. OpenAI is now scrambling to pivot toward the sticky, reliable enterprise dollars, while Anthropic, through necessity and success, is learning to operate in the public spotlight. The company’s deliberate, long-term focus on responsible scaling and security has not slowed its growth; rather, it has provided the credibility necessary to secure the foundational enterprise contracts and massive compute deals that fuel its continued ascent.</p>
<div class="lb-trans"><p>Anthropic 和 OpenAI 之间的竞争正在改变这两家公司的形态。OpenAI 现在正急于转向稳定、可靠的企业收入，而 Anthropic 则因必要性和成功而学会在公众视野中运作。该公司对负责任的扩展和安全性的深思熟虑、长期关注并没有减缓其增长；相反，这为其获得基础企业合同和巨额计算交易提供了必要的信誉，推动了其持续上升。</p>
</div>

Anthropic 的安全优先策略正在定义下一个万亿美元级别的 AI 市场