超过十年来，软件公司凭借向企业销售长期订阅来建立可预测且高利润的 SaaS 收入结构。彭博报导的『SaaSpocalypse』已经因 AI 可能颠覆甚至取代既有定价模式而已经重创多家上市 SaaS 公司的市值。当亚马逊云端部门宣布开发可自动化例如销售这类业务的 AI 代理后，追踪软件的 ETF 下跌超过 4%，部分单只股票下跌约 9%，显示市场正在对订阅模式风险进行广泛重新定价。
投资者与运营者现在担忧的，不仅是 AI 替代风险，还有合同层面的锁定力正在被削弱。据 Andreessen Horowitz 的医疗投资者 Julie Yoo 所言，以前服务医疗机构的初创企业通常追求多年合约，如今多改为一年期，甚至按月计费，让医院在 AI 工具快速更迭时可更快切换。一些医院现在会同时开展多个试点，将不同 AI 产品「并列比较」后再做决定，反映出对灵活性和降低转换成本的高需求。
这种转变在其他行业同样可见。已营运 12 年的法律技术平台 Filevine，面对 unicorn 如 Harvey 和 Legora 的竞争，正在加入 agentic AI 层，使律师可直接查询既有系统，并加快企业客户的上线速度。Menlo Ventures 的 Derek Xiao 指出，企业采购越来越像自消费市场：员工先在周末试用 Lovable 或 Anthropic 的 Claude 等工具，再把它们带入工作场景；Forrester 的 Lisa Singer 则说企业不再愿意签订三年 AI 价格合同，因为三年后 AI 成本谁也难预测，而风险投资也观察到销售周期正从几个月缩短到几周，利润波动与流失风险也因此上升。
For more than a decade, software firms built durable, high-margin SaaS revenue using enterprise long-term subscriptions. Bloomberg says the so-called “SaaSpocalypse” has already slammed the market values of many public SaaS companies, because AI may disrupt or replace their existing pricing models. When Amazon’s cloud unit announced an AI agent to automate tasks such as sales, the ETF tracking software fell more than 4%, with some individual stocks down about 9%, showing broad market repricing of subscription-model risk.
Investors and operators now worry not only about AI replacement risk but also weakening lock-in at the contract level. According to Andreessen Horowitz healthcare investor Julie Yoo, startups serving healthcare previously pursued multi-year contracts; many now offer one-year agreements, even month-to-month terms, so hospitals can switch faster as AI tools evolve quickly. Some hospitals are now running multiple pilots at once, “baking off” competing AI products before committing, reflecting high demand for flexibility and lower switching costs.
The shift is visible in other sectors too. Twelve-year-old legal-tech platform Filevine, facing unicorn competitors Harvey and Legora, is adding an agentic AI layer so lawyers can directly query existing systems and onboard enterprise customers faster. Derek Xiao of Menlo Ventures says enterprise purchasing now resembles consumer demand: employees try tools like Lovable or Anthropic’s Claude on weekends and then carry them into work. Lisa Singer of Forrester says enterprises are no longer willing to sign fixed three-year AI pricing, because no one can predict AI costs three years out; venture investors also observe sales cycles shrinking from months to weeks, with higher volatility and churn risk. (Key numbers: 12)