由Google Cloud赞助
选择您的第一个基因AI 使用案例
要从基因AI开始，首先关注能够改善人类信息体验的领域。
证监会将销售商商业增长规模的揭示归档，同时提出了企业买家如何分辨AI支出是否还清的更宽泛的问题。
5 读取
仅通过 Getty 映像
Anthropic本周的IPO备案突出了企业对前沿AI模型日益增长的需求，但企业是否从这些投资中获得可计量的价值仍然是企业领袖的一个关键问题。
公司报告2026年上半年收入163亿美元。 其企业客户基础也迅速扩大：截至9月30日，根据证监会的备案，Anthropic拥有11,900多客户，在过去12个月中，每个客户产生至少10万美元的收入，比一年前的950多岁有所增加。
对部署前沿AI的企业来说，挑战在于确定应用技术的全额成本是否合理。 这意味着超越模型或订阅价格来核算一体化、人力监督、持续维护和供应商锁定的风险。
Anthropic的数据显示其技术支出增长多快，但仅靠客户数和提供者收入并不能显示企业是否在获得该支出的回报。
企业对AI的需求加速
AI模型的进步和能够写出代码，分析信息并实现多步骤业务流程自动化的agents的出现，正在扩大企业在实验之外如何使用技术。
Jeet Pattanaik、SAP API程序管理者、Glokal AI公司创始人和CTO(Glokal AI公司是一家企业AI销售商)称这些预付款，加上更容易购买，正在帮助推动收养。
"简单的理由是，模型现在已经足够做真正的活了","Pattanaik 称。 现有的云合同还可以使企业更容易获得AI，并将其使用范围扩大到试点项目之外，he 称。
然而，由于公司将AI整合到更多的工作流程中，单靠使用并不能揭示技术提供的价值有多大。 雇员可能经常使用人工智能工具，而不使其工作流程更快或更有效。 影响取决于AI是降低成本，节省时间，提高生产力还是提高效果。
企业AI支出既反映了对技术的信心，也反映了董事会和首席执行官要求更快行动的压力，据全球科技和服务公司Concentrix全球产品组织副总裁Kathryn Harrison称。 "这不应该被解读为证明公司已经了解了经济学","Harrison 称。 企业还需要投资于业务变革，使雇员和技术能够有效合作，她就是称。
带来新的成本和依赖性
随着企业将AI推入更多的工作流程，将技术投入工作的成本可能变得更加难以管理。 在初步试验期间看来可以管理的开支随着部署的扩大而会增加。
Tim Dol，世代集团首席数字官和Precocity的首席执行官，技术顾问，称企业经常低估了规模化部署的成本，特别是当他们使用昂贵的前沿模型进行日常任务时。
人类监督新增对这些费用，因为员工必须检查AI生成的输出的准确性和实用性，多尔称。
企业在数据不齐全，系统连接不畅的情况下，也可以面临挑战。
“大多数组织并没有真正考虑的另外一件事是数据准备状态和连接系统。”
Harrison 称公司还必须为雇员培训和工人学习新系统的时间编列预算。 随着AI技术的改变，企业可能需要重新审视或重建工作流程，从而在初始部署之后增加成本。
Anthropic的备案也说明了AI的经济学对于供应商及其客户的不同。 称公司在企业工作流程和产品中扩展了Claude的使用，有助于推动收入增长。 它还描述了一种周期，即降低每个任务的费用鼓励更多的使用，这可以导致更广泛的采用。 对Anthropic来说，这种扩张可以产生额外的收入。 但对于企业来说，问题是增加AI使用的好处是否证明增加的费用是合理的。
模型定价是计算的一个因素。 随着供应商提高效率并采用价格较低的模式，企业可能能够更廉价地承担某些工作量。 "厨师模型是好消息，但他们很少降低总账单","Pattanaik 称。 "当一些东西变得更便宜时，人们会更多地使用。 所以，每个任务的成本都比较低，但总体开支却增加了".
对于企业购买者来说，相关的衡量标准不仅仅是每个令牌或订阅的价格。 相对于其交付的价值，是部署和维护AI驱动的工作流程的总成本。 这些费用可以包括部署规模、从整合和监督到持续维护等隐蔽费用。
缩放AI还可以使交换提供者更加困难。 一个使用Claude进行有限实验的企业可以相对容易地尝试另一个模型。 但围绕特定模型构建应用程序，编码工作流程或agents的企业可能会面临巨大的切换成本。
Pattanaik 称企业应询问它们是否可以在三个月内将人工智能工作量转移到另一个供应商。 如果转换需要更长的时间，公司可能会过于依赖其目前的供应商。 内部保持提示、工作流程和测试系统可以使这些工具更容易适应不同的模式。
"规范你如何使用AI，而不是你如何围绕它构建","Pattanaik 称。
真正的测试是企业价值
Anthropic的IPO备案在业务规模上给出了数字，但企业领导面临不同的计算：他们的AI投资是否改善了业绩，足以证明成本的合理性。
“AI供应商通常对使用收费，而不是客户所取得的成果。 这让企业负责确定成功是什么样子
Harrison 称公司应当将完成某项任务的成本与AI与之前流程的成本进行比较，包括员工花费的时间审查和纠正AI生成的工作。 它们还应衡量AI是否改善了业务成果，例如降低服务成本、加快完成任务或改善客户经验和收入。
"公司还必须评估整个工作流程，因为如果AI在别处创造出额外的工作，一步中的节省就可以消失"(Harrison 称。
Anthropic将市场规模的形成归档为显示，但证明这些投资的收益仍然是购买企业的责任。
Kinza Yasar涵盖AI和新兴技术TechTarget，重点是道德、企业的采纳、治理和商业战略。 在进入新闻工作之前，她担任了IT和网络支持角色，使她从系统层面看待企业技术的构建，部署和管理。
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
The SEC filing reveals the scale of the vendor’s commercial growth while raising a broader question of how enterprise buyers can tell if AI spending is paying off.
5 Min Read
Just_Super via Getty Images
Anthropic's IPO filing this week highlights growing enterprise demand for frontier AI models, but whether businesses are getting measurable value from those investments remains a key question for enterprise leaders.
The company reported $16.3 billion in revenue for the first half of 2026. Its enterprise customer base has also expanded rapidly: As of Sept. 30, Anthropic had more than 11,900 customers generating at least $100,000 in revenue each during the previous 12 months, up from more than 950 a year earlier, according to the SEC filing.
For businesses deploying frontier AI, the challenge is determining whether the benefits justify the full cost of deploying the technology. That means looking beyond model or subscription prices to account for integration, human oversight, ongoing maintenance and the risk of vendor lock-in.
Anthropic's figures show how quickly spending on its technology is growing, but customer counts and provider revenue alone don't reveal whether enterprises are earning a return on that spending.
Enterprise demand for AI accelerates
Advances in AI models and the emergence of agents capable of writing code, analyzing information and automating multistep business processes are expanding how enterprises can use the technology beyond experimentation.
Jeet Pattanaik, SAP API program manager and founder and CTO of Glokal AI, an enterprise AI vendor, said these advances, combined with easier purchasing, are helping drive adoption.
"The simple reason is that the models are now good enough to do real work," Pattanaik said. Existing cloud contracts can also make it easier for enterprises to access AI and expand its use beyond pilot projects, he said.
However, as companies integrate AI into more workflows, usage alone doesn't reveal how much value the technology delivers. Employees might regularly use AI tools without making their workflows faster or more effective. The impact depends on whether AI reduces costs, saves time, boosts productivity or improves results.
Enterprise AI spending reflects both confidence in the technology and pressure from boards and CEOs to move faster, according to Kathryn Harrison, global vice president, global product organization at Concentrix, a global technology and services company. "It shouldn't be read as proof that companies have figured out the economics," Harrison said. Enterprises also need to invest in the operational changes that enable employees and technology to work together effectively, she said.
Deeper adoption brings new costs and dependencies
As enterprises move AI into more workflows, the costs of putting the technology to work can become harder to manage. Expenses that seem manageable during initial experiments can grow as deployments expand.
Tim Doll, chief digital officer of Generational Group and CEO of Precocity, a tech consultancy, said businesses often underestimate the expense of scaling deployments, especially when they use expensive frontier models for routine tasks.
Human oversight adds to those costs because employees must check the accuracy and usefulness of AI-generated outputs, Doll said.
Enterprises can also face challenges when their data isn't ready, and their systems aren't well connected.
"The other thing that most organizations don't really factor in is the readiness of their data and connecting systems," he said.
Harrison said companies must also budget for employee training and the time workers spend learning new systems. As AI technology changes, enterprises might need to revisit or rebuild workflows, adding costs beyond the initial deployment.
Anthropic's filing also illustrates how the economics of AI can differ for providers and their customers. The company says expanding Claude's use across enterprise workflows and products helps drive revenue growth. It also describes a cycle in which lower per-task costs encourage greater usage, which can lead to broader adoption. For Anthropic, that expansion can generate additional revenue. But for businesses, the question is whether the benefits of increased AI use justify the added expense.
Model pricing is one factor in that calculation. As providers improve efficiency and introduce lower-priced models, enterprises might be able to run certain workloads more cheaply. "Cheaper models are good news, but they rarely lower the total bill," Pattanaik said. "When something gets cheaper, people use it more. So, each task costs less, but overall spending goes up."
For enterprise buyers, the relevant metric isn't simply the price per token or subscription. It's the total cost of deploying and maintaining an AI-enabled workflow compared with the value it delivers. Those expenses can include hidden costs that emerge as deployments scale, from integration and oversight to ongoing maintenance.
Scaling AI can also make it harder to switch providers. An enterprise using Claude for a limited experiment can try another model relatively easily. But a business that builds applications, coding workflows or agents around a particular model might face significant switching costs.
Pattanaik said enterprises should ask whether they could move their AI workloads to another provider within three months. If switching were to take longer, the company could become too dependent on its current provider. Keeping prompts, workflows and testing systems in-house can make it easier to adapt those tools to a different model.
"Standardize how you use AI, but not how you build around it," Pattanaik said.
The real test is enterprise value
Anthropic's IPO filing puts a number on the scale of its business, but enterprise leaders face a different calculation: whether their AI investments improve performance enough to justify the costs.
“AI providers generally charge for usage rather than the results customers achieve. That leaves businesses responsible for establishing what success looks like and measuring whether a deployment delivers it," Doll said.
Harrison said companies should compare the cost of completing a task with AI against the cost of the previous process, including the time employees spend reviewing and correcting AI-generated work. They should also measure whether AI improves business results, such as reducing service costs, speeding up task completion or improving customer experience and revenue.
"Companies must also assess the entire workflow, because savings in one step can disappear if AI creates extra work elsewhere," Harrison said.
Anthropic's filing shows the scale of the market taking shape, but proving the return on those investments remains the responsibility of the businesses buying in.
Kinza Yasar covers AI and emerging technology for TechTarget, with a focus on ethics, enterprise adoption, governance and business strategy. Before moving into journalism, she worked in IT and network support roles, giving her a systems-level perspective on how enterprise technologies are built, deployed and managed.
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.