Micro1 hits $500M gross run rate as demand for AI training data grows

In short: Micro1, a startup that supplies training data for AI systems, has reportedly reached a $500 million gross annual run rate as demand surges.

What's going on

Companies building AI are spending heavily on training data, which is the example material an AI system learns from (like flashcards used to study for a test). That demand is helping a group of “data labeling” startups grow quickly.

TechCrunch reports that Micro1, a four-year-old company, increased its gross annual run rate from $100 million to $500 million in the past eight months, according to a person familiar with the company. “Gross run rate” is a rough estimate of yearly revenue based on current pace, not a final audited number.

Micro1 reportedly keeps about 60% to 70% of that gross figure after paying contractors, putting its net annual run rate between $150 million and $200 million. Like rivals, it uses contract workers with specific expertise, including doctors, lawyers, and scientists, to help create and check AI training data.

Micro1 is also doing more “synthetic data,” which means data made by software instead of people (like auto written descriptions of what happens in a video). Some of its datasets can be sold to more than one customer, which can raise margins. Selling the same training data to multiple buyers has also drawn criticism in the industry, especially around whether some companies provide data to Chinese AI developers.

What to watch

Researchers cited by TechCrunch suggest future AI budgets for data could become as large as budgets for computing power, meaning this market may keep expanding. Micro1’s founder has said the company does not sell data to Chinese model makers, and questions about where training data goes may get more attention as these firms grow.

Source: TechCrunch AI