AI data startup Micro1 has 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 (20/08).

Micro1, a four-year-old startup, keeps about 60% to 70% of its gross annual run rate. This puts its net annual run rate at between $150 million and $200 million.

The growth comes as leading AI laboratories and corporations show strong demand for unique AI training data, supporting rapid expansion among data-labeling startups.

Micro1 remains behind competitors such as Mercor, which reached $2 billion in gross annualized revenue this summer, and Handshake, which reached $1 billion earlier this year.

AI Training Data Demand Fuels Startup Growth

The strong demand for AI training data is supporting multiple companies that provide data for AI development.

Like other companies in the sector, Micro1 works with domain experts, including doctors, lawyers, and scientists, on a contract basis.

Some researchers have suggested that future AI spending on data could become comparable to spending on computing power.

Micro1 is also seeing its contract sizes increase at a faster rate and expects its margins to expand over time.

Micro1 Expands Synthetic and Off-the-Shelf Data

Micro1 is increasingly producing synthetic data without human involvement. This includes creating automated descriptions of video content.

Some of the data produced by the company can also be sold to multiple customers. A person familiar with Micro1’s finances told TechCrunch that margins for this “off-the-shelf” data can reach 80% to 90%.

The company’s ability to sell the same datasets to multiple customers has become controversial, particularly over concerns about providing data to Chinese AI developers.

Data Sales to Chinese AI Developers Spark Controversy

Critics have argued that distributing off-the-shelf data to Chinese AI developers could help their models become as powerful as leading U.S. models.

Micro1 founder Ali Ansari said last month on X that the company does not sell its data to Chinese model makers.

“Some human data companies work with foreign adversaries. [A]nd the results show today in Kimi K3. We believe it’s shameful to claim American AI dominance desires while selling millions worth of data to countries that we are in adversarial competition with.”

Micro1 Shifted From AI Recruiting to Data Labeling

Micro1 initially operated as an AI recruiting startup, similar to Mercor.

Ansari decided to expand into data labeling after noticing that clients were using Micro1’s AI platform to evaluate and recruit engineers for annotation.

The company has also had its experts evaluate AI model outputs, a process known as reinforcement learning gyms.

Micro1 is building a robotics pre-training dataset by having hundreds of generalists record interactions with everyday objects in their homes.

Micro1 raised its Series A at a $500 million valuation last September. TechCrunch understands that the startup may have recently raised another funding round at a significantly higher valuation.

Micro1 did not respond to a request for comment.

PHOTO: MICRO1

This article was created with AI assistance.

We make every effort to ensure the accuracy of our content, some information may be incorrect or outdated. Please let us know of any corrections at [email protected].