The ARR Illusion: How Some AI Startups Inflate Revenue Metrics

In a typical SaaS company, ARR answers a simple question: if the current subscription base stayed exactly the same—no new customers, no churn—how much recurring revenue would the company generate over the next 12 months?

The metric becomes more complicated in early‑stage AI startups, where revenue often mixes subscriptions, usage‑based pricing, pilot projects, and enterprise contracts.

Tactic #1: Contracted ARR (CARR)

One increasingly debated approach is contracted ARR, sometimes called committed ARR.

Instead of counting only revenue from active subscriptions, some startups include the value of contracts that:

• Have been signed but have not yet started
• Include revenue ramp‑ups in future years
• Depend on product features that are still being delivered

In certain cases, startups even calculate ARR based on the highest future value of a contract from day one, rather than the revenue currently being billed.

For example, if a contract grows from $100,000 this year to $300,000 in year three, a CARR calculation might count the $300,000 immediately—even though that revenue may not materialize for years. This can produce a large gap between headline ARR and the company’s present‑day revenue base.

Critics argue that this approach transforms ARR from a snapshot of recurring revenue into something closer to total contracted potential value.

Tactic #2: Annualized Run‑Rate Revenue

Another common tactic is using revenue run rate.

Run‑rate calculations take revenue from a short recent period—often a single month or quarter—and multiply it to project a full year. For instance:

• Monthly revenue × 12
• Quarterly revenue × 4

This can provide a quick estimate of scale, but it assumes the current pace will remain constant.

That assumption can be risky because early‑stage companies often experience:

• Large one‑time contracts
• Seasonal spikes
• promotional usage bursts
• pilot programs that do not renew

Run‑rate projections therefore describe a possible trajectory, not stable recurring revenue. If growth slows or usage drops, the projected annual number can quickly diverge from reality.

Why These Metrics Can Be Misleading

The problem is not necessarily that these metrics exist—it’s that they can blur fundamentally different concepts.

A single startup might report one large headline number labeled “ARR” that actually mixes three different things:

• Current recurring subscription revenue
• Signed but not‑yet‑activated contract value
• Projections based on recent revenue bursts

When those categories are merged into a single figure, outsiders may assume the company already has a stable revenue base at that scale.

Yet the company’s GAAP revenue—the revenue recognized under accounting standards—may be much smaller.

Under the accounting rule ASC 606, companies recognize revenue only when goods or services are delivered and the performance obligation is satisfied. Signing a contract alone does not create immediate revenue recognition.

This difference means a startup can announce a large ARR figure while its recognized accounting revenue, cash collections, or active subscriptions are significantly lower.

Why Investors and Media Sometimes Accept It

Despite the controversy, these inflated‑looking metrics continue to appear in fundraising decks and press coverage. Several incentives help explain why.

First, venture capital markets often value future growth potential more than current financial precision. Aggressive forward‑looking metrics can reinforce a narrative that a startup is quickly becoming a category leader.

Second, venture investors benefit when portfolio companies appear to be accelerating faster than competitors. A headline like “$100M ARR in two years” can help attract follow‑on funding and increase valuations.

Third, media coverage tends to reward dramatic growth stories. Articles announcing record‑breaking ARR milestones are easier to publish than deep analyses of how the number was calculated.

Finally, AI products can genuinely scale usage extremely quickly, which makes large annualized numbers seem plausible—even if the underlying revenue is volatile.

The Metric Comparison That Actually Matters

To interpret startup revenue claims accurately, it helps to separate four different measures:

ARR
A non‑GAAP snapshot of current recurring subscription revenue from active customers.

Contracted ARR (CARR)
A broader estimate that may include signed deals that haven’t started or revenue scheduled to ramp up in the future.

Annualized run rate
A projection created by extrapolating revenue from a recent period over a full year.

GAAP revenue
Revenue recognized under accounting standards when contractual obligations are actually fulfilled.

Each number can be useful—but they answer different questions.

What Investors Are Increasingly Asking

As scrutiny grows, investors are asking startups to clarify how their ARR is calculated. Key questions often include:

• How much of ARR is live and currently billed?
• How much represents signed but not yet activated contracts?
• What portion is usage‑based rather than fixed subscription revenue?
• How does the ARR figure reconcile with GAAP‑recognized revenue?

These distinctions help separate a real recurring revenue base from a projection or pipeline estimate.

The Bottom Line

ARR has long been a useful shorthand for measuring SaaS growth. But in the AI startup boom, the metric’s flexibility has allowed some companies to present numbers that look far larger than their realized business.

Understanding the difference between true recurring revenue, contracted future revenue, and annualized projections is increasingly essential for interpreting the headline growth claims emerging from the AI startup ecosystem.