文 | 高恒
半年营收超过去年全年三倍，净亏损却高达807.48亿美元。10月26日，美国人工智能公司Anthropic, PBC（以下简称“Anthropic”）向美国证券交易委员会（SEC）公开提交首次公开发行（IPO）注册文件，披露了这份反差鲜明的成绩单。招股书显示，2026年上半年，公司实现营收162.86亿美元，约为上年同期的14倍，营业亏损却从21.35亿美元收窄至6.93亿美元。
巨额净亏损主要来自融资工具的公允价值变动，并不能直接反映公司的实际经营表现。随着企业客户增加、模型使用需求扩大，Anthropic的盈利能力正在改善。但面对OpenAI、谷歌等竞争对手持续推出新模型、调整价格，以及已经签订的巨额长期算力采购合同，公司能否维持收入增长，并将其转化为稳定利润，仍是此次上市需要面对的重要考验。
01 企业客户激增，半年营收超过去年全年三倍
Anthropic的营收在过去三年持续快速增长。
招股书显示，2023年，Anthropic实现营收4100万美元，2024年增长至3.86亿美元，2025年进一步达到45.85亿美元。
进入2026年，这一增长势头进一步加快。上半年，公司实现营收162.86亿美元，约为2025年同期11.63亿美元的14倍，超过去年全年营收的三倍。其中，第一季度营收47.31亿美元，第二季度进一步增长至115.55亿美元。
企业客户对人工智能模型的采购和使用，是推动Anthropic收入增长的重要来源。
与单纯依靠固定订阅费的收费方式不同，Anthropic同时提供按使用量计费和订阅服务。企业可以通过应用程序编程接口（API）或云计算平台调用Claude模型，并根据实际使用量支付费用。
招股书显示，2026年上半年，Anthropic按使用量计费的收入达到130.64亿美元，占总营收约80%，2025年同期则为9.83亿美元；订阅收入由上年同期的1.80亿美元增长至32.22亿美元，占总营收约20%。
两类收入均实现大幅增长，但按使用量计费的业务贡献了约80%的同比收入增量。这类业务主要根据模型处理的数据量等指标收费，客户调用模型越多，通常需要支付的费用也越高。
随着企业将Claude应用于更多实际业务，Anthropic的大客户数量也在快速增加。
招股书披露，截至2026年9月30日，过去12个月为Anthropic贡献超过10万美元收入的客户已超过1.19万家，而2025年同期仅有950多家。其中，贡献超过100万美元收入的客户超过1900家，约250家客户贡献的收入达到1000万美元以上。
从收入规模和客户数量来看，Anthropic的商业化已经从早期市场拓展进入企业客户持续扩大采购的阶段。其中，代码开发是其形成竞争优势的重要领域。
Anthropic推出的Claude Code可以协助开发人员完成代码生成、修改和调试等任务，逐步进入企业软件开发流程。相较于普通对话应用，代码开发的使用需求更加具体，企业也能够通过开发效率评估相关工具的实际价值。
美国风险投资机构Menlo Ventures于2025年12月发布的《2025年企业生成式人工智能发展状况》（2025: The State of Generative AI in the Enterprise）报告指出，基于美国企业人工智能使用调查及市场模型估算，Anthropic在企业大模型API支出市场的份额达到40%，高于OpenAI的27%和谷歌的21%。
从市场变化来看，Anthropic的估算份额由2023年的12%提升至2025年的40%，同期OpenAI则由50%下降至27%。在代码开发相关模型市场，Anthropic的估算份额达到54%，OpenAI为21%。Menlo Ventures认为，Claude Code的普及是推动Anthropic份额提升的重要原因之一。
另一家美国企业费用管理平台Ramp的支付数据，也反映了两家公司在企业市场的竞争变化。
Ramp于10月14日发布的人工智能指数研究（Ramp AI Index）显示，2026年9月，在其平台企业客户中，43.8%向Anthropic支付了相关费用，高于OpenAI的40.6%。
但从当月的增长情况来看，两家公司有所不同。与8月相比，Anthropic企业付费采用率提高0.25个百分点，OpenAI则提高1.12个百分点，双方差距有所缩小。
Ramp首席经济学家Ara Kharazian在报告中指出，随着企业AI付费采用率提高，新增客户增长正在放缓，未来市场增长将更加依赖现有客户扩大使用规模，而非单纯增加采购企业的数量。
企业客户扩张推动Anthropic收入迅速增长，但提供模型服务同样需要持续投入计算资源。收入规模不断扩大之后，公司的盈利表现却呈现出另一番景象。
02 净亏损超过800亿美元，营业亏损却大幅收窄
招股书显示，2023年至2025年，Anthropic分别净亏损15.31亿美元、83.09亿美元和419.65亿美元。2026年上半年，净亏损进一步扩大至807.48亿美元，远高于同期162.86亿美元的营收。
造成巨额亏损的主要原因并非模型研发和日常经营支出，而是融资工具的会计估值变化。
2026年上半年，Anthropic因可转换票据、衍生负债及未来股权简单协议（SAFE）等融资工具的公允价值变动，确认损失723.88亿美元，占同期净亏损的近九成。此外，公司还确认其他净费用87.57亿美元，主要涉及融资相关协议的初始公允价值确认及后续估值变化。
这些损失与Anthropic的融资结构密切相关。公司此前通过可转换票据等方式筹集资金，部分工具的价值随企业估值变化而调整。当Anthropic估值提高时，相应金融负债的公允价值也可能上升，并在利润表中形成损失，但并不意味着公司同期实际支付了等额现金。
巨额账面亏损之外，Anthropic的营业利润却出现了明显改善。
2025年上半年，Anthropic营业亏损21.35亿美元，营业亏损率约为183.6%。2026年同期，营业亏损收窄至6.93亿美元，营业亏损率降至4.3%。
从季度来看，今年第一季度，Anthropic仍录得16.67亿美元营业亏损，第二季度则实现9.74亿美元营业利润，为招股书披露期间首次单季度营业利润转正。
这一变化主要来自收入增长速度超过营业费用增速。
招股书显示，2026年上半年，Anthropic营业费用合计169.79亿美元，2025年同期为32.98亿美元，约增长4.1倍；而同期营收增长约13倍。收入增加151.23亿美元，营业费用增加136.81亿美元，两者差额对应营业亏损同比减少的14.42亿美元。
其中，算力及基础设施仍是Anthropic最大的费用项目。
2025年上半年，公司算力及基础设施费用为23.34亿美元，2026年同期增加至89.45亿美元，同比增长283%，甚至超过2025年全年的73.26亿美元。
但随着营收迅速扩大，这项费用占营收的比重已从2025年上半年的约200.7%下降至54.9%。
Anthropic在招股书中解释，算力费用增长主要受到此前签订的云服务合同陆续交付计算资源的影响。相关费用通常在算力可供使用时确认，而非签订合同当时确认。随着收入增长，提前锁定的算力成本在营收中的占比相应下降。
按照招股书附注8对营业费用的职能分类，Anthropic的毛利率也出现明显改善。
2026年上半年，公司营收成本为23.80亿美元，据此计算的毛利率约为85.4%，2025年同期约为12.4%。同期，Anthropic研发费用达到97.41亿美元。
营收成本主要包括直接用于服务客户的计算资源及相关费用，而研发费用则涵盖模型研究、开发等投入。两者与前述按费用性质列示的算力及基础设施费用属于不同分类方式，不能重复计算。
除了算力，人员和销售费用也在快速增长。
2026年上半年，Anthropic人员费用由上年同期的5.38亿美元增至31.52亿美元，其中约22亿美元为股权激励费用；其他经营及管理费用由3.06亿美元增至17.14亿美元。
增长更明显的是销售、营销及合作伙伴费用，从上年同期的1.20亿美元增至31.68亿美元。公司在招股书中将其中约27亿美元的同比增量归因于渠道合作伙伴费用增长。
这一变化与Anthropic的销售模式有关。2026年上半年，公司约48%的营收通过云服务商市场渠道实现。相关平台帮助Anthropic接触企业客户，公司则需要按照协议支付合作费用。
现金流方面，2026年上半年，Anthropic经营活动现金流净额达到39.29亿美元，而上年同期为净流出15.13亿美元。
招股书显示，这一变化除了经营表现改善，还受到应付款项增加、预付费用减少及递延收入增长等因素影响。同期，公司应收款项也随业务扩张明显增加。
整体来看，收入增长快于营业费用增长，推动Anthropic营业亏损明显收窄，经营现金流也有所改善。但随着竞争对手持续推出新模型，客户需求和价格变化，将直接影响这种经营改善能否持续。
03 模型竞争加剧，长期算力投入仍是考验
招股书显示，Anthropic将OpenAI和谷歌列为面向开发者及企业客户市场的主要竞争对手。公司同时指出，部分竞争者拥有更雄厚的资金实力、技术资源以及更充足的算力供应。
其中，OpenAI正在持续拓展企业客户和代码开发市场，谷歌则同时拥有Gemini模型、云计算平台和自研计算芯片。对于Anthropic来说，竞争不仅发生在模型能力和产品价格上，还涉及客户渠道、计算资源以及持续研发的资金投入。
这种竞争已经反映在企业更换模型供应商的行为上。
Ramp于10月6日发布的研究显示，截至2026年9月，其样本中更换AI模型供应商的企业比例，按三个月移动平均计算已达到8%，创下该项统计的新高。
报告指出，2025年5月Anthropic推出Claude Opus 4和Sonnet 4后，曾吸引更多企业转向其模型。此后更换供应商的情况逐步减少，但2026年6月OpenAI推出新模型并下调价格，再次改变了市场竞争格局。
Ramp在10月14日发布的研究进一步显示，截至10月11日的一周，其Token支出管理样本的Token使用量较9月13日当周增长52.4%，同期支出仅增长23%。
在该样本中，企业使用的Token数量增长明显快于实际支出。模型价格、使用结构和计算效率的变化，使大模型公司未必能够获得与使用量同比例增长的收入。
企业能够根据模型性能和价格调整供应商，也使Anthropic需要不断推出具有竞争力的新产品。
招股书披露，2026年第三季度，公司超过80%的收入来自此前六个月内发布的模型版本。Anthropic表示，新模型发布后，大部分客户会逐步转向性能更强的版本。
新模型不断推出，需要持续投入研发和训练资源，而算力采购通常需要提前安排。
招股书显示，截至2026年6月30日，Anthropic未来固定或最低采购承诺约为2656亿美元，相关合同期限为1至20年。公司另行披露，与亚马逊云科技的相关算力采购协议最低金额至少为1100亿美元，与谷歌云的相关采购承诺至少约1145亿美元。相关协议属于不同披露范围，金额不能简单累加。
长期采购能够保障未来模型研发和客户使用所需的计算资源，但部分合同设有最低付款义务，不会随着客户使用量减少而同步下降。
相比之下，Anthropic的大部分收入来自按实际使用量计费的模型服务，部分企业客户没有长期采购承诺，可以根据模型性能、价格和自身需求调整供应商。
如果市场竞争使客户转向其他模型，或者模型价格下降导致收入增速放缓，Anthropic仍需要承担已经约定的部分算力支出。
这种竞争与合作关系，在大型云服务商身上表现得尤为明显。
亚马逊、谷歌和微软既为Anthropic提供计算资源，又通过相关云平台销售Claude服务，同时在部分人工智能业务上与Anthropic存在竞争。
以谷歌为例，其既拥有Gemini模型和云计算基础设施，能够向Anthropic提供算力，也可以直接向企业客户销售自身的AI模型。对于Anthropic而言，云平台既是拓展客户和获取算力的重要渠道，相关采购价格与合作费用又直接影响公司的经营成本。
为了改善计算资源供应和长期成本，Anthropic正在推进“Anthropic Direct”基础设施模式，通过直接租赁计算芯片、部署专用计算资源等方式，增加自身对算力的调配能力。
这种模式能够减少部分对传统云服务采购方式的依赖，但也意味着公司需要承担设备租赁及其他基础设施费用，并提前安排相应资金。
从企业客户扩张到营业利润转正，Anthropic已经展现出规模化经营带来的盈利改善。但面对OpenAI、谷歌等竞争对手持续推出新模型、调整价格，以及长期算力合同陆续执行，公司还需要保持足够的客户需求，并提高计算资源的使用效率。
对于即将登陆资本市场的Anthropic来说，收入增长已经有了实际业绩支撑，而这些收入能否在持续的模型竞争和算力投入中转化为稳定利润，将是下一阶段更重要的考验。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Man, Queen, Kosheng.
Half-year collections were three times higher than the previous year, while net losses amounted to $80,748 million. On 26 October, the United States artificial intelligence firm Anthropic, the PBC (hereinafter “Anthropic”), publicly submitted to the United States Securities and Exchange Commission (SEC) the first publicly released (IPO) registration document, disclosing this contrasting report card. According to the equity book, in the first half of 2026, companies earned $16,286 million, or 14 times more than during the same period the previous year, while operating losses narrowed from $2,135 million to $693 million.
Large net losses arise mainly from changes in the fair value of financing instruments and do not directly reflect the actual performance of companies. The profitability of Anthropic is improving as business clients increase and demand for model use expands. However, in the face of the continued introduction of new models, price adjustments by competitors such as Open AI, Google, and the huge long-term competitive procurement contracts that have been signed, the ability of companies to sustain income growth and convert it into stable profits remains an important test for listing.
01 Business client surge, six months more than three times more than last year
Anthropic revenue collection has continued to grow rapidly over the past three years.
According to the book, in 2023 Anthropic earned $41 million, rising to $386 million in 2024 and further reaching $4.585 billion in 2025.
In 2026, the momentum for this growth accelerated further. In the first half of the year, companies received $16,286 million, about 14 times the amount of $1,163 million received during the same period in 2025, three times the amount collected throughout the year. Of this amount, $4,731 million was collected in the first quarter and further increased to $11,555 million in the second quarter.
The purchase and use of artificial intelligence models by business clients is an important source of income growth for Anthropic.
Unlike a fee that relies solely on fixed subscription fees, Anthropic provides both usage-based and subscription services. Enterprises can call the Claude model through the Application Programming Interface (API) or the Cloud Calculator Platform and pay paid based on actual usage.
In the first half of 2026, the income of Anthropic on the basis of usage amounted to $13,064 million, representing about 80 per cent of the total, compared with $983 million in the same period in 2025; the income from subscriptions increased from $180 million during the same period of the previous year to $3,222 million, or about 20 per cent of the total.
Both types of income showed significant growth, but the usage-based operations contributed about 80 per cent of the same increase in income. Such operations are charged mainly on the basis of indicators such as the volume of data processed by the model, and the more the model is called upon by the client, the higher the costs usually charged.
As businesses apply Claude to more practical operations, the number of large clients of Anthropic is growing rapidly.
The equity book disclosed that, as at 30 September 2026, more than 119,000 clients had contributed more than $100,000 in income to Anthropic over the past 12 months, compared to more than 950 in the same period in 2025. Of these, more than 1,900 clients contributed more than $1 million of income, and some 250 clients contributed more than $10 million.
In terms of size of income and number of clients, the commercialization of Anthropic has moved from early market expansion to the stage of continuous expansion of procurement by business clients. Of these, code development is an important area of their competitive advantage.
Claude Code, launched by Anthropic, can assist developers in their task of coding, modifying and debugging and gradually entering the enterprise software development process. Compared to normal dialogue applications, the use needs for code development are more specific, and enterprises are able to assess the practical value of relevant tools through the development of efficiency.
In December 2025, the United States venture capitalist Menlo Ventures reported the Enterprise Generating Artificial Intelligence Development in 2025 (2025: The State of General AI in the Enterprise) that Anthony ' s share in the corporate LLMAPI expenditure market was 40 per cent, up from 27 per cent in Open AI and 21 per cent in Google, based on the United States Enterprise Artificial Intelligence Use Survey and market models.
In terms of market changes, the estimated share of Anthropic rose from 12 per cent in 2023 to 40 per cent in 2025, while Open AI fell from 50 per cent to 27 per cent over the same period. In the market for code development-related models, the estimated share of Anthropic reached 54 per cent and that of Open AI 21 per cent. According to Menlo Ventures, the popularization of Claude Code is one of the important reasons behind the increase in the number of Anthropic shares.
The payment data of Ramp, another United States corporate cost management platform, also reflect the evolution of competition between the two companies in the enterprise market.
Ramp ' s artificial intelligence index study (Ramp AI Index), dated 14 October, showed that in September 2026, 43.8 per cent of its platform business clients paid related fees to Anthropic, which was 40.6 per cent higher than Open AI.
However, in view of the increase in the current month, the two companies differ. In comparison with August, the rate of adoption of paid by Anthropic Enterprise increased by 0.25 percentage points, while Open AI increased by 1.12 percentage points, narrowing the gap between the two sides.
In his report, Ara Kharazian, Chief Economist of Ramp, noted that growth in new clients was slowing as the adoption rate of the enterprise AZpaid increased, and that future market growth would rely more on existing clients to expand their use than simply increasing the number of purchasing firms.
The expansion of business clients has driven the rapid growth of Anthropic revenues, but the provision of modelling services also requires sustained investment in computing resources. The growing size of revenues is followed by a different picture of corporate profitability.
02 Net losses in excess of $80 billion, while operating losses were significantly narrowed
According to the equity book, from 2023 to 2025, Anthropic had a net loss of $1,531 million, $8309 million and $41,965 million, respectively. In the first half of 2026, the net loss further increased to $80,748 million, significantly higher than the $16,286 million received during the same period.
The major cause of the large losses was not model development and day-to-day operating expenses, but changes in the accounting valuation of financing instruments.
In the first half of 2026, Anthropic recognized losses of $72,388 million, or nearly 90 per cent of the net loss for the same period, due to changes in the fair value of financing instruments such as convertible instruments, derivative liabilities and simple future equity agreements (SAFE). In addition, the company recognized other net costs of $8,757 million, mainly relating to the initial fair value recognition of financing-related agreements and subsequent valuation changes.
These losses are closely related to the financing structure of Anthropic. Companies had previously raised funds through, for example, transferable instruments, and the value of some of the instruments had been adjusted to reflect changes in the valuation of the enterprise. When the Anthropic valuation increases, the fair value of the corresponding financial liability may also increase and be lost in the profit statement, but this does not mean that the company actually paid the equivalent amount of cash during the same period.
In addition to the large book losses, Anthropic ' s operating profits have improved markedly.
In the first half of 2025, Anthropic had an operating deficit of $2,135 million and an operating deficit of about 183.6 per cent. During the same period in 2026, the operating losses narrowed to $693 million and the operating losses decreased to 4.3 per cent.
On a quarterly basis, in the first quarter of this year, Anthropic recorded a business loss of $1,667 million, while in the second quarter it realized a business profit of $974 million, which was reversed for the first single-quarter period of the equity book disclosure period.
This change is mainly due to the fact that income is growing faster than operating costs.
According to the book, in the first half of 2026, the operating costs of Anthropic totalled $16,979 million, compared to $3,298 million in the same period in 2025, representing a growth of about 4.1 per cent, compared to a growth of about 13 times during the same period. Income increased by $15,123 million and operating costs increased by $13,681 million, reflecting a decrease of $1,442 million in operating losses.
Of these, computing and infrastructure remain the largest cost items for Anthropic.
In the first half of 2025, corporate capacity and infrastructure costs amounted to $2,334 million, rising to $8,945 million in the same period in 2026, an increase of 283 per cent over the same period, or even more than $7,326 million throughout 2025.
However, with the rapid expansion of revenue collection, the share of this cost in revenue collection fell from about 20.7 per cent in the first half of 2025 to 54.9 per cent.
In its equity book, Anthropic explained that the increase in capacity costs was mainly affected by the successive delivery of computing resources under the previously concluded cloud service contract. The associated costs are usually recognized when the available resources are available, rather than when the contract is concluded. As income increases, the share of early-targeted calculus costs in revenue has declined accordingly.
Anthropic also showed a marked improvement in its Māori rate, according to the functional classification of operating costs in note 8 to the equity book.
In the first half of 2026, the company ' s collection cost was $2,380 million, resulting in a gross domestic product ratio of approximately 85.4 per cent and 12.4 per cent for the same period in 2025. During the same period, the cost of research and development in Anthropic amounted to $9.741 billion.
Cost recovery includes mainly computing resources and associated costs that are directly used to serve clients, while research and development costs cover inputs such as model research and development. The two are classified in a different way from the cost nature of the foregoing and infrastructure costs, and cannot be double-counted.
In addition to capacity, the cost of personnel and sales is growing rapidly.
In the first half of 2026, personnel costs for Anthropic increased from $538 million during the same period of the previous year to $3,152 million, of which some $2.2 billion represented equity incentive costs; and other operating and management costs increased from $306 million to $1,714 million.
More notably, the increase was in sales, marketing and partner costs, from $120 million in the same period of the previous year to $3,168 million. In its equity book, the company attributed some $2.7 billion of this increase to the increase in the costs of the channel partners.
This change is related to Anthropic's sales pattern. In the first half of 2026, approximately 48 per cent of the company ' s revenues were channelled through the cloud service provider market. The platform helps Anthropic to reach out to business clients, and the company is required to pay for the cooperation as agreed.
In terms of cash flows, in the first half of 2026, net cash flows from operating activities in Anthropic amounted to $3.929 billion, compared to $1.513 billion in net outflows during the same period of the previous year.
The equity book shows that this change, in addition to improved business performance, is also affected by increased accounts payable, reduced use of advance paid and increased deferred income. During the same period, corporate receivables also increased significantly as business expanded.
Overall, income grew faster than operating costs, contributing to a marked narrowing of the operating losses of Anthropic and improving operating cash flows. However, as competitors continue to introduce new models, changes in client demand and prices will have a direct impact on the sustainability of such business improvements.
03 Increased competition for models and long-term calculus inputs still test
According to the book, Anthropic lists Open AI and Google as major competitors in the market for developers and business customers. At the same time, companies pointed out that some competitors had greater financial power, technical resources and a better calculus supply.
Among them, Open AI is continuously expanding the business customer and code development market, while Google has the Gemini model, cloud computing platform and self-study computing chip. For Anthropic, competition takes place not only in modelling capacity and product prices, but also in customer channels, computing resources and financial inputs for ongoing research and development.
This competition has been reflected in the behaviour of enterprises in changing model suppliers.
The Ramp study of publish, dated 6 October, showed that, as of September 2026, the proportion of businesses that had replaced AI model suppliers in sample had reached an average of 8 per cent on a three-month moving basis, creating a new high in that statistics.
The report notes that, following the introduction of Claude Opus 4 and Sonnet 4 by Anthropic in May 2025, more businesses were attracted to their models. Since then, there has been a gradual reduction in the replacement of suppliers, but in June 2026 Open AI introduced a new model and reduced prices, which again changed the pattern of competition in the market.
Ramp ' s study of publish on 14 October further showed that, as of the week of 11 October, Token ' s expenditure management Z111Z had a 52.4 per cent increase in its usage over the week of 13 September, while expenditure had increased by only 23 per cent over the same period.
In this sample, the number of Tokens used by enterprises increased significantly faster than the actual expenditure. Changes in model prices, use structures and computational efficiency have not necessarily enabled LLM companies to generate income on the same scale as their usage.
The ability of enterprises to adjust suppliers to model performance and prices also requires Anthropic to continuously introduce competitive new products.
The equity book disclosed that in the third quarter of 2026, more than 80 per cent of the company ' s revenues were derived from the model version of publish in the previous six months. Anthropic states that, following the new model publish, most clients will gradually shift to a more performing version.
The roll-out of the new model will require sustained investment in research and development and training resources, while the acquisition of computing power will usually require advance arrangements.
According to the requisition, as at 30 June 2026, Anthropic ' s future fixed or minimum procurement commitment was approximately $26.56 billion, with the relevant contract duration ranging from 1 to 20 years. The company separately disclosed that the minimum value of the power purchase agreement relating to Amazon cloud technology was at least US$ 110 billion and that the purchase commitments related to Google Clouds were at least US$ 114.5 billion. The agreement is subject to different disclosure and the amount cannot simply be added up.
Long-term procurement can guarantee future model development and client use of the computing resources required, but some contracts have minimum payment obligations and do not decline simultaneously with fewer customer usages.
In contrast, the bulk of Anthropic ' s income is derived from model services based on actual usage, and some corporate clients have no long-term purchase commitment to adjust their suppliers to model performance, prices and their own needs.
Anthropic still has to bear part of the calculus expenditures that have been agreed upon if competition in the market leads to a shift of customers to other models, or if the decline in model prices slows the pace of income growth.
This competition and cooperation is particularly evident among large cloud service providers.
Amazon, Google and Microsoft both provide computing resources for Anthropic and sell Claude services through associated cloud platforms, while competing with Anthropic in part of artificial intelligence operations.
Google, for example, has both the Gemini model and cloud computing infrastructure, which can provide calculators to Anthropic and sell its own AI models directly to business clients. For Anthropic, the cloud platform is an important channel for reaching out to clients and gaining leverage, and the associated procurement prices and cooperation costs directly affect the operating costs of companies.
In order to improve the calculation of resource availability and long-term costs, Anthropic is moving forward with the “Anthropic Direct” infrastructure model to increase its ability to deploy its own calculus by directly leasing chips and deploying dedicated computing resources.
This model can reduce some reliance on traditional cloud services procurement, but it also means that companies have to bear the cost of leasing equipment and other infrastructure and pre-position the corresponding funds.
From expansion of business clients to business profits, Anthropic has demonstrated the improvement in profitability resulting from scale operations. However, in the face of the continued introduction of new models, price adjustments and the implementation of long-term computing contracts by competitors such as Open AI and Google, companies also need to maintain sufficient customer demand and improve the efficiency of the use of computing resources.
For Anthropic, who is about to land in capital markets, the growth of income has been underpinned by real performance, and whether these revenues translate into stable profits in sustained model competition and calculus inputs will be a more important test in the next phase.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.