The burgeoning artificial intelligence sector, a hotbed of innovation and intense venture capital interest, is currently grappling with a significant challenge to its financial transparency. A recent public outcry, spearheaded by Scott Stevenson, co-founder and CEO of the legal AI startup Spellbook, has cast a harsh spotlight on what he terms a "huge scam" involving the widespread inflation of revenue figures announced by some AI companies. Stevenson’s direct allegations, shared on social media platform X, contend that many AI startups are achieving "revenue records" through the deployment of "dishonest metrics," a practice he claims is tacitly supported by major investment funds keen on generating positive public relations and misleading media outlets.

The Genesis of a Concern: A Shifting Metric Landscape

Stevenson is far from the first to highlight the potential for manipulation surrounding Annual Recurring Revenue (ARR), a metric historically revered in the software-as-a-service (SaaS) industry. For years, ARR has served as a critical barometer, summing up the annual revenue generated from active customers under long-term contracts. However, its application, particularly within the fast-paced and often speculative AI market, appears to have stretched its original definition beyond recognition. Previous reports in financial media and numerous social media discussions have already alluded to various "shenanigans" concerning ARR calculations. Yet, Stevenson’s specific and pointed tweet resonated deeply within the AI startup ecosystem, sparking widespread debate, drawing over 200 reshares and comments from prominent investors, founders, and leading to several follow-up news stories.

The widespread reaction suggests that Stevenson articulated a sentiment many in the industry harbored but hesitated to voice publicly. Jack Newton, co-founder and CEO of the legal tech firm Clio, commended Stevenson for "highlighting some of what you might describe as bad behavior," acknowledging the much-needed awareness his post generated. Newton further referenced an explanatory post by Garry Tan of Y Combinator, a prominent startup accelerator, which detailed proper revenue metrics, underscoring the urgency of the issue. A subsequent investigation, involving interviews with over a dozen founders, investors, and startup finance professionals, largely corroborated Stevenson’s claims. Many sources, speaking anonymously due to the sensitivity of the topic, confirmed that the public declaration of fudged ARR figures is indeed a prevalent practice among startups, and crucially, that investors are often fully aware of these exaggerations.

Decoding Recurring Revenue: ARR vs. CARR

At the heart of the controversy lies a fundamental distinction between established financial reporting and newer, more flexible interpretations. The primary obfuscation tactic identified is the substitution of "contracted ARR," sometimes referred to as "committed ARR" (CARR), for true ARR. While both metrics aim to project future revenue, their underlying assumptions and reliability diverge significantly.

ARR, in its traditional sense, represents the annualized value of all active subscription contracts at a given point in time. It is a robust indicator of a company’s predictable, recurring income stream, crucial for valuation models in the SaaS sector where stable, long-term customer relationships are paramount. It reflects revenue from customers who are actively using and paying for a service. Crucially, Generally Accepted Accounting Principles (GAAP) primarily focus on historical, already-collected revenue, meaning ARR, as a forward-looking metric, is not formally audited by accountants. However, its widespread acceptance in the tech investment world has made it a de facto standard for assessing growth potential.

CARR, on the other hand, expands upon the ARR concept by including revenue from signed customer contracts that have not yet gone live or been fully implemented. As one anonymous investor noted, "For sure they are reporting CARR" as ARR, acknowledging the competitive pressure: "When one startup does it in a category, it is hard not to do it yourself just to keep up." While CARR can be a useful internal metric for tracking future pipeline, it is inherently "squishier" than traditional ARR. Bessemer Venture Partners (BVP), a leading venture capital firm, acknowledged CARR in a 2021 blog post on scaling, advising that startups should adjust CARR figures to account for expected customer churn and "downsell" – instances where customers reduce their service commitment.

The critical vulnerability of CARR lies in counting revenue before product implementation. Lengthy or problematic onboarding processes can lead to clients canceling during trial periods or even before services commence, meaning a significant portion of that "contracted" revenue may never materialize. This creates a disconnect between announced figures and actual cash flow. One venture capitalist revealed witnessing companies where CARR exceeded ARR by as much as 70%, with a substantial portion of that contracted revenue ultimately failing to materialize. There are documented instances of high-profile enterprise startups publicly announcing they surpassed $100 million in ARR, when only a fraction originated from currently paying customers, with the remainder tied to contracts awaiting deployment, some with potentially lengthy implementation timelines. One former employee at a startup candidly described a situation where a substantial, year-long free pilot was counted as ARR, a practice known to the company’s board, including a VC from a major fund. This board was also aware of the customer’s ability to cancel before the full contract amount was paid.

The ease with which CARR can be "gamed" further complicates matters. Without realistic adjustments for churn and downsell, a startup could offer steep discounts for the initial years of a multi-year contract, then count the entire contract value as CARR (or ARR), even if customers are unlikely to remain at higher prices in later years. Ross McNairn, co-founder and CEO of legal AI startup Wordsmith, echoed Stevenson’s concerns, noting, "I’ve heard all sorts of anecdotes as well. I speak to VCs all the time. They’re like, ‘There are some choppy, choppy standards out.’" While some discrepancies might be less extreme, such as marketing materials claiming $50 million in ARR when the actual figure is $42 million, the underlying issue persists. Sources indicated that investors, despite having access to accurate internal books, often tolerate these public misrepresentations, especially in the context of the AI boom where rapid growth is expected, viewing an $8 million gap as a "rounding error" that the company will quickly grow into.

The Other "ARR": Annualized Run-Rate Revenue

Adding another layer of complexity to the revenue reporting landscape is the use of "annualized run-rate revenue," another metric often abbreviated as "ARR." This version of ARR extrapolates a short-term revenue haul – be it from a single quarter, month, week, or even a day – over a full 12-month period. While potentially useful for very early-stage startups to show momentum, it is inherently speculative and susceptible to significant fluctuations.

For many AI companies, whose business models increasingly rely on usage-based or outcome-driven pricing rather than fixed, predictable subscriptions, annualized run-rate revenue can be particularly misleading. Revenue in these models is not "locked in" by long-term contracts in the same way traditional SaaS ARR is. A surge in usage during a specific week might lead to an impressive annualized figure, but it offers no guarantee of sustained performance. This type of "ARR" can dramatically inflate perceived growth without reflecting underlying stability or customer commitment.

The AI Hype Cycle and Investor Incentives

The phenomenon of revenue overstatement is not entirely new to the startup world, but its prevalence and aggressiveness have intensified dramatically amid the current AI hype cycle. The unprecedented valuations commanded by AI startups have created a powerful incentive structure. Michael Marks, a founding managing partner at Celesta Capital, observed that "The valuations have gotten higher, and so the incentives are stronger to do it."

The venture capital ecosystem thrives on identifying and nurturing "unicorns" – companies valued at over $1 billion. In the age of AI, the expectation for growth has become extraordinarily ambitious. Hemant Taneja, CEO and managing director of General Catalyst, articulated this sentiment on the 20VC podcast, stating that traditional growth trajectories like "1 to 3 to 9 to 27 [million in ARR] is not interesting." Instead, he declared, "You got to go like 1 to 20 to 100." This immense pressure to demonstrate hyper-growth pushes both founders and investors to embrace, or at least tolerate, more aggressive public reporting strategies.

Stevenson explicitly points to the complicity of venture capitalists: "There are definitely VCs in on this because they’re incentivized to create a narrative that they have runaway winners. They’re incentivized to get press coverage for their companies." Newton, whose company Clio achieved a $5 billion valuation, concurs, noting, "We see some investors looking the other way when their own companies are inflating numbers because it makes them look good from the outside in."

Market Distortions and the Call for Transparency

This selective blindness by investors serves a strategic purpose. By overlooking public pronouncements of inflated ARR, VCs effectively engage in a "kingmaking" strategy, elevating their portfolio companies to perceived market leaders. High reported revenue attracts not only further investment but also top talent and key customers who seek to align with perceived winners. As one anonymous VC put it, "Investors can’t call it out. Everyone has a company monetizing CARR as ARR."

However, this collective silence has broader market implications. It distorts competition, potentially misallocating capital to companies with less robust underlying financials. It also creates an unrealistic benchmark for other startups, forcing them into a "race to the bottom" where integrity might be sacrificed for perceived growth. For those intimately familiar with the industry, the rapid ascent of some AI startups to $100 million ARR within a few years of launch often "just feels fake," as Alex Cohen, co-founder and CEO of health AI startup Hello Patient, articulated. "You read the headlines and you’re like, ‘I don’t believe it.’"

Not all founders, however, are willing to participate in this trend. Many prioritize clear and transparent financial reporting, understanding that public markets, should their companies pursue an IPO, will scrutinize traditional ARR metrics rigorously. McNairn of Wordsmith, recalling the painful market correction of 2022 that forced many startups to justify their valuations, emphasizes the long-term dangers of such practices. "I think it is short-sighted, and I think that when you do things like that for a short-term gain, you’re overinflating already crazy high multiples," he warned. "I think it’s super bad hygiene, and it’s going to come back and bite you." The increasing scrutiny on AI startup revenue practices signals a growing demand for greater financial integrity and a potential reckoning for companies that prioritize narrative over verifiable performance.