The world of AI startups is booming, with unprecedented valuations and rapid growth seemingly around every corner. Yet, behind the impressive headlines, a quiet but significant concern has emerged: are these “revenue records” as solid as they appear? A recent social media post by Scott Stevenson, co-founder and CEO of legal AI startup Spellbook, sparked a crucial conversation, calling out what he termed a “huge scam” in how some AI companies report their revenue figures.
Stevenson’s claims struck a nerve, prompting over 200 reshares and comments from prominent investors and founders. He highlighted how a “dishonest metric” is being used, with major funds reportedly supporting this trend and misleading journalists for favorable PR. This isn’t an entirely new issue, but the current AI hype seems to have amplified the practice, pushing traditional reporting standards to their limits.
The Blurry Lines of “ARR”
At the heart of the debate is the manipulation of Annual Recurring Revenue (ARR), a metric historically used to denote the predictable annual income from active customers under signed contracts. This standard was established during the cloud era to reflect the total value of sealed sales, often multi-year agreements, where usage and payments are metered over time. It represents revenue that is reliably expected, often from already active services.
However, many AI companies are now presenting a different figure, often interchangeably called ARR, which is actually Contracted Annual Recurring Revenue (CARR). CARR broadens the definition by including revenue from signed customers who haven’t even been onboarded or begun using the product yet. While potentially useful for tracking future growth, this metric is considerably “squishier” and far more susceptible to inflation.
Why the Numbers Don’t Add Up
The core issue with reporting CARR as ARR is its inherent unreliability. A significant chunk of this “contracted” revenue may never materialize if product implementation is lengthy, complex, or goes awry, leading clients to cancel before any payments are collected. One venture capitalist revealed instances where CARR was an astonishing 70% higher than actual ARR, with a substantial portion of that contracted revenue never materializing.
Beyond this, some companies have been observed including free pilot programs, even those lasting a year, within their reported ARR figures. This means revenue from an eventual paying portion of a contract is counted well before any money changes hands, and crucially, before the customer commits to paying. Such practices allow startups to present a hyper-accelerated growth trajectory that doesn’t align with actual cash flow or solidified customer relationships.
The Pressure Cooker: Why Startups & VCs Play Along
The intense competitive landscape of the AI sector and the expectation of exponential growth create immense pressure on startups to showcase impressive revenue figures. As one investor noted, “When one startup does it in a category, it is hard not to do it yourself just to keep up.” VCs themselves contribute to this environment, often seeking companies that can jump from “$1 million to $20 million to $100 million” in ARR far faster than traditional projections.
Many investors, while aware of these discrepancies, find themselves in a complex position. There’s a strong incentive to back “runaway winners” and generate positive press for their portfolio companies. This narrative helps attract top talent, secure more customers, and ultimately drive higher valuations. As one VC put it, “Everyone has a company monetizing CARR as ARR,” making it difficult to call out the practice without undermining their own investments.
Another, less common, but equally opaque metric often conflated with traditional ARR is “annualized run-rate revenue.” This figure extrapolates current revenue over 12 months based on a short period’s (e.g., a month or even a day) performance. For many AI companies that charge based on usage or outcomes, this can be highly misleading, as revenue isn’t locked into predictable, fixed contracts and can fluctuate wildly.
The Long-Term Consequences of Inflated Metrics
While publicly inflating ARR might offer short-term gains, many insiders are growing wary. The perception that reported figures are “fake” or unbelievable erodes trust within the ecosystem. As one founder remarked, “To everyone who’s inside, it just feels fake. You read the headlines and you’re like, ‘I don’t believe it.'”
The danger is that these inflated numbers create an unsustainable valuation bubble, setting unrealistic expectations for future performance. Founders who prioritize transparency understand that public markets measure software companies on true ARR, not CARR, and that exaggerating growth can lead to significant hurdles down the line. As Ross McNairn, co-founder of Wordsmith, succinctly put it, “I think it is short-sighted… it’s super bad hygiene, and it’s going to come back and bite you.”
Source: TechCrunch – AI
