Run Rate

Run Rate

The run rate extrapolates a short period of time to a full year: revenue of 10 million euros in a month results in a run rate of 120 million euros a year. It doesn't show actual earnings, but a projection assuming everything continues at the same pace.

A company brings in 10 million euros in a month. Multiply that by twelve and you get 120 million euros a year. This exact projection is called the run rate. It doesn’t describe what was actually earned, but what would add up over a year if things continued exactly as they have recently. Instead of a month, a quarter, i.e. three months, is often used, multiplied by four. The figure is thus a snapshot stretched out to twelve months.

Why young companies love calculating this

For a company that has operated steadily for twenty years, the run rate is pretty uninteresting. There, the real annual revenue is already known from the last annual report. The figure becomes interesting for companies that are growing very fast. A look back at the past year would massively underestimate their current size.

This is especially true for AI companies of recent years. If a provider had barely any revenue in January and is bringing in 100 million euros a month by December, the annual revenue says little. A run rate of 1.2 billion euros describes the current pace better. That’s why run-rate figures almost always show up in headlines about OpenAI, Anthropic, or Nvidia's customers.

But that’s exactly why the figure is also popular when you want to impress. It sounds bigger than actual revenue and can be dressed up with just one good month. Investors know this and ask how stable the underlying month was. For readers of business news, that’s the single most important follow-up question.

The math behind it, and what it assumes

The formula is simple: revenue for a period divided by its length, multiplied by twelve months. That turns 30 million euros in a quarter into a run rate of 120 million euros. Sometimes it’s not revenue that gets extrapolated, but profit, the number of customers, or a data center’s power consumption. The principle stays the same.

The calculation lives or dies by one assumption: that the chosen period is typical. An ice cream vendor with a run rate based on July would obviously be dishonest. With software, it’s less obvious, but just as possible. One large one-off order or a marketing push can heavily distort a single month.

Related, but not identical, is ARR, Annual Recurring Revenue. It only counts recurring income, such as ongoing subscriptions. One-time payments are excluded. An ARR is therefore more reliable than a plain run rate, which includes everything that came in during the period. Because both terms are often mixed up in press releases, it’s worth checking exactly what is meant.

How to correctly read run-rate figures in reports

You’ll most often encounter the run rate in reports about tech companies without a stock listing. Such companies aren’t required to publish audited figures. So they voluntarily state a metric, and usually the most flattering one. Phrases like “reaches a run rate of one billion dollars” don’t mean that a billion has actually come in.

Companies also calculate this way internally. A team checks after a quarter whether it’s on track for its annual target. It works the same way on the cost side: anyone spending 5 million euros a month on computing power has a cost run rate of 60 million euros. For AI companies with expensive graphics cards, this is a figure to take seriously.

Three questions help when reading such figures. Which period was extrapolated, a month or a quarter? Was that period normal or especially good? And is it about revenue or profit? After all, a high revenue run rate says nothing about whether a company is actually profitable.

Subscribe free. Unsubscribe the second it sucks.

High-signal news across AI, business, UX, and tech. Every morning.