Revenue Run Rate

The revenue run rate is an extrapolation. You take the earnings of a short period, say a month, and project them out to a full year. A company that brings in ten million euros in December has a run rate of 120 million euros per year. But it never actually earned those 120 million. The figure only describes a pace, much like a car’s speedometer: 100 kilometers per hour doesn’t mean you’ve already driven 100 kilometers. Precisely for this reason, the run rate is both useful and dangerous.

Why young tech companies love using it

A normal annual financial statement looks backward. It shows what happened over the past twelve months. For a company that doubles every few months, this view is almost worthless. The figures from the previous year describe a company that no longer exists in that form. The run rate puts the current state front and center.

That’s why this metric turns up especially often among AI companies and other startups. When OpenAI or Anthropic release reports of billion-dollar revenues, they almost always mean a run rate. Investors value such companies as a multiple of this figure. A jump in the run rate from two to four billion can double the estimated company valuation without a single fiscal year having been completed.

There is also a solid, factual reason for the metric. Anyone selling subscriptions has predictable, recurring revenue. A customer who pays 20 euros a month will, with high probability, do so again next month. For such business models, the extrapolation is considerably more reliable than for one-off sales.

Projected from one month to twelve

The calculation itself is simple. Monthly revenue times twelve, quarterly revenue times four, weekly revenue times 52. There is no fixed rule for which period forms the basis. And that is exactly where the first bit of leeway lies: a company can pick its best month and build the run rate from that.

The second bit of leeway lies in the underlying assumption. The calculation presumes that nothing changes. No customer cancels, no competitor cuts prices, no holiday season comes to an end. A ski rental business with a February run rate of twelve million euros will never actually reach that sum over the year. Its business is seasonal, making the extrapolation meaningless.

It’s also important to distinguish this from profit. The run rate only measures what comes in, not what’s left over. Many AI companies have high run rates and still post deeply negative figures, because data centers and staff devour even more money. A related metric is ARR, annual recurring revenue. It explicitly counts only recurring subscription income and leaves out one-off revenues.

Reading the figure correctly when it appears in the news

In press releases and business news, the run rate usually appears in phrases like “projected on an annual basis” or “annualized.” These words are the signal. As soon as they show up, it’s not actually earned money but a forecast built from a single snapshot value.

Two questions help with putting it into context. First: what period does the figure come from? A run rate based on a single strong month is weaker than one based on an entire quarter. Second: how stable is the revenue? Subscriptions are more reliable than one-off orders or fixed-term contracts with only a few major clients.

You’ll encounter this principle outside the world of finance, too. A student job paying 300 euros a month corresponds to a run rate of 3,600 euros a year. If the job disappears during the holidays, the math no longer holds. The run rate is therefore not a trick, but a legitimate tool with one clear condition: it’s only as good as the assumption that everything keeps going the way it is.