The Letter VCs Are Quietly Deleting from ARR

风投们正在悄悄从 ARR 中删掉的一个字母

Startups are reporting revenue they haven’t earned yet. VCs know it. Investors are cheering anyway. We’ve seen this movie. 初创公司正在报告他们尚未赚取的收入。风投们心知肚明，投资者们却依然在欢呼。这一幕我们似曾相识。

You’re evaluating an AI startup. The pitch deck shows $100 million in ARR. The growth curve is parabolic. The deck says they signed $100 million. What it doesn’t say is that $70 million of that is “committed ARR”: contracts signed but not yet invoiced, customers who haven’t deployed yet, pilots that count toward the number if they convert. Subtract the gap and you’ve got $30 million in actual recognized revenue hiding under a $100 million headline. This is the ARR inflation playbook, and it’s running at full speed right now. 你正在评估一家人工智能初创公司。融资推介演示文稿（Pitch Deck）显示其年度经常性收入（ARR）为 1 亿美元。增长曲线呈抛物线状。文稿称他们签署了 1 亿美元的合同。但它没说的是，其中 7000 万美元是“承诺 ARR”（Committed ARR）：即已签署但尚未开票的合同、尚未部署的客户，以及如果转化成功则计入总额的试点项目。扣除这些水分，你得到的实际确认收入仅为 3000 万美元，却被隐藏在 1 亿美元的标题之下。这就是 ARR 通胀的剧本，而且目前正全速运转。

The Trick Has a Name

这个把戏有个名字

“CARR” stands for Contracted or Committed ARR. It’s a legitimate concept. In industries where revenue accrues slowly after contract signing (healthcare AI deployments, energy optimization platforms, multi-year enterprise integrations), the gap between signature and recognition can legitimately take months or years to close. Reporting CARR alongside ARR, properly labeled, is defensible. That’s not what’s happening. What’s happening is simpler: founders strip the “C” and just call it ARR. “CARR”代表合同或承诺的 ARR。这是一个合法的概念。在合同签署后收入增长缓慢的行业（如医疗 AI 部署、能源优化平台、多年期企业集成），从签署到确认收入之间确实可能需要数月甚至数年。如果标注清晰，将 CARR 与 ARR 一起报告是站得住脚的。但现在的情况并非如此。发生的事情更简单：创始人去掉了“C”，直接称其为 ARR。

One VC told TechCrunch the gap between CARR and actual ARR can run as high as 70% [1]. In some confirmed cases the spread is 3-5x. Another investor said flatly: “For sure they are reporting CARR as ARR” [1]. The article indicates that the investor community is not only aware, but many are actively complicit. The logic follows its own warped rationality. When one startup in a category inflates, the others have to follow to stay competitive for talent and headlines. “When one startup does it in a category, it is hard not to do it yourself just to keep up,” as one investor put it [1]. 一位风投告诉 TechCrunch，CARR 与实际 ARR 之间的差距最高可达 70% [1]。在一些已证实的案例中，差距甚至达到 3 到 5 倍。另一位投资者直言不讳地说：“他们肯定是在把 CARR 当作 ARR 报告” [1]。文章指出，投资者群体不仅知情，许多人甚至积极参与其中。这种逻辑遵循着一种扭曲的合理性。当同一赛道的一家初创公司开始注水时，其他公司为了在人才竞争和媒体曝光中保持竞争力，不得不跟进。正如一位投资者所言：“当赛道里有一家公司这么做时，为了不掉队，你很难不效仿” [1]。

Spellbook CEO Scott Stevenson, one of the few willing to call this in public, described the practice as a “huge scam,” adding that major VC funds are not just watching it happen but actively supporting the narrative [2]. The metric isn’t tracking business reality. It’s tracking a story being told for the benefit of people who need the story to be true. Spellbook 的首席执行官 Scott Stevenson 是少数愿意公开指出这一点的人之一，他将这种做法描述为一场“巨大的骗局”，并补充说大型风投基金不仅在旁观，而且在积极支持这种叙事 [2]。这一指标追踪的不是商业现实，而是为了那些需要这个故事成真的人所编造的叙事。

We Have Seen This Before

我们以前见过这一幕

If you were in the SaaS world circa 2021-2022, this will feel familiar. At the peak of the zero-interest-rate era, growth-at-all-costs was the gospel. Companies reported “ARR” using definitions that would have made their accountants wince: annualizing a single good month, counting trials, counting LOIs, counting anything that could be plausibly labeled recurring. The metric became a narrative tool, not a financial one. 如果你在 2021-2022 年左右身处 SaaS 行业，这会让你感到熟悉。在零利率时代的巅峰时期，“不惜一切代价增长”是金科玉律。公司报告“ARR”时使用的定义会让会计师们皱眉：将一个表现良好的月份年化、计入试用期、计入意向书（LOI），以及任何可以勉强贴上“经常性”标签的东西。该指标变成了一种叙事工具，而非财务工具。

The crash was instructive. When rates rose and growth-multiple valuations compressed, the distance between reported metrics and cash reality became unforgiving. Companies that had raised at 40x ARR on inflated numbers found themselves underwater fast. The write-downs from venture portfolios in 2022 and 2023 were staggering. 随后的崩盘很有教育意义。当利率上升、增长倍数估值压缩时，报告指标与现金现实之间的差距变得残酷。那些基于虚高数字以 40 倍 ARR 融资的公司很快发现自己陷入了困境。2022 年和 2023 年风投组合的减记规模令人震惊。

Go back further to the dot-com era and the dynamic is even starker. Eyeballs. Page views. Registered users. Each bubble generates its own vanity metric that sounds like a financial number but behaves like a PR number. The AI cycle’s version is CARR-as-ARR: precise enough to sound real, slippery enough to hide the gap. The pattern is always the same: a new category, a new metric, and a crowd of people with aligned incentives telling themselves the old rules don’t apply. 回溯到互联网泡沫时代，这种动态更为明显。眼球数、页面浏览量、注册用户数。每一次泡沫都会产生自己的虚荣指标，听起来像财务数字，表现起来却像公关数字。AI 周期的版本就是“将 CARR 当作 ARR”：它精确到听起来很真实，又模糊到足以掩盖差距。模式总是相同的：一个新的赛道、一个新的指标，以及一群利益一致的人告诉自己，旧规则不再适用。

Why This Is Actually a Ponzi Problem

为什么这实际上是一个庞氏骗局问题

Run the actual math. An AI startup raises at, say, 20x its reported $100 million ARR. Valuation: $2 billion. But if actual ARR is $30 million, the real multiple is 67x: a number most rational investors wouldn’t accept if they saw it plainly. The gap has to close in one of two ways: either the revenue materializes (the CARR converts, the contracts activate, the pilots become paid customers) or the next fundraise arrives before anyone asks hard questions. 算算实际的账。一家 AI 初创公司以其报告的 1 亿美元 ARR 的 20 倍进行融资。估值：20 亿美元。但如果实际 ARR 是 3000 万美元，那么真实的倍数是 67 倍：如果理性投资者能看清这一点，大多数人是不会接受的。这个差距必须通过两种方式之一来弥补：要么收入真正实现（CARR 转化、合同激活、试点项目成为付费客户），要么在有人提出尖锐问题之前完成下一轮融资。

The next round keeps the story alive. The round after that does too. The round after that is either an IPO, an acquisition, or a reckoning. When markets are liquid and multiples are expanding, the gap can stay hidden indefinitely. The new money funds the old story. But when the music stops (rising rates, a public market correction, a category reset), the companies sitting on inflated ARR find themselves without chairs. 下一轮融资让故事继续。再下一轮也是如此。再往后要么是 IPO、被收购，要么就是清算。当市场流动性充裕且倍数扩张时，差距可以无限期隐藏。新资金为旧故事买单。但当音乐停止时（利率上升、公开市场调整、赛道重置），那些坐在虚高 ARR 上的公司就会发现自己无处藏身。

The investors who got in early will have returned their funds by the time the reckoning arrives. The investors who got in at round D or E on a $2 billion valuation backed by $30 million in real revenue will not. The employees who took lower salaries for equity will not. This is not hyperbole. This is the documented playbook of every prior cycle, replayed with different vocabulary. 当清算时刻到来时，早期进入的投资者早已收回了资金。而那些在 20 亿美元估值、仅有 3000 万美元真实收入支撑的 D 轮或 E 轮进入的投资者则不会。那些为了股权而接受低薪的员工也不会。这不是夸大其词。这是每一个过往周期中都有据可查的剧本，只是换了一套词汇重新上演。

What to Actually Look At

真正需要关注什么

If you’re evaluating AI companies (as an investor, an acquirer, a PE firm doing diligence, or a potential enterprise customer checking vendor stability), here’s what the ARR number won’t tell you: 如果你正在评估 AI 公司（作为投资者、收购方、进行尽职调查的私募股权公司，或检查供应商稳定性的潜在企业客户），以下是 ARR 数字无法告诉你的信息：

Cash in the bank and burn rate. If a company has $100 million in ARR and $40 million in cash with a $6 million monthly burn, the story is different than if they’re sitting on $200 million. ARR doesn’t pay rent. Cash does. 银行存款和烧钱率。 如果一家公司有 1 亿美元 ARR，账上有 4000 万美元现金，每月烧钱 600 万美元，这与账上有 2 亿美元的情况完全不同。ARR 支付不了房租，现金才可以。
Revenue recognition policy. What accounting standard are they using? When do they book revenue: at contract signing, at deployment, at invoice, at collection? A company that recognizes revenue at signature on multi-year contracts is telling a very different story than one that recognizes monthly on delivery. This requires asking directly. The answer tells you which ARR definition you’re working with. 收入确认政策。 他们使用什么会计准则？他们何时入账收入：合同签署时、部署时、开票时还是收款时？一家在多年期合同签署时就确认收入的公司，与一家按月交付确认收入的公司，讲述的是完全不同的故事。这需要直接询问。答案会告诉你你所面对的是哪种 ARR 定义。
Customer concentration. “$100 million in ARR” is a very different number if three customers represent 60% of it. Realization risk and churn risk are both concentrated. 客户集中度。 如果 1 亿美元 ARR 中有 60% 来自三个客户，那么这个数字的含义就完全不同了。实现风险和流失风险都高度集中。
Logo churn vs. dollar churn. The churned customer count understates dollar exposure. The customer who leaves quietly is rarely the small one. 客户流失率与金额流失率。 流失的客户数量会低估金额风险。悄悄离开的客户很少是小客户。
Pilot-to-paid conversion rate. AI adoption timelines are long. Pilots run 90 to 180 days. Not all convert. 试点到付费的转化率。 AI 的采用周期很长。试点项目通常持续 90 到 180 天。并非所有试点都能转化。