The immense and constant demand for distinctive AI training data from leading laboratories and corporations is fueling a massive surge for a group of data-labeling startups.
Micro1, a four-year-old startup, is one of these rapidly expanding enterprises, having increased its gross annual run rate from $100 million to $500 million over the past eight months, according to a source familiar with the company. Similar to its industry counterparts that contract domain specialists such as doctors, lawyers, and scientists, Micro1 retains approximately 60% to 70% of this revenue, placing its net annual run rate between $150 million and $200 million.
While Micro1 still trails competitors like Mercor (which achieved $2 billion in gross annualized revenue this summer) and Handshake (which reached $1 billion earlier this year), the startup’s revenue expansion demonstrates that there is more than sufficient demand to support multiple companies supplying AI training data.
This rapid growth is expected to persist, with some researchers suggesting that future AI expenditures on data could eventually equal those on computational power.
This outlook is favorable for Micro1, which is experiencing an accelerated increase in contract sizes and anticipates expanding its profit margins over time. The startup is increasingly generating synthetic data without human intervention, for instance, by creating automated descriptions of video content. Furthermore, some of the data it produces can be sold to multiple clients, driving gross margins for this “off-the-shelf” data as high as 80% to 90%, a source familiar with the startup’s financial details informed TechCrunch.
The practice of selling identical datasets to multiple clients has recently sparked controversy, with critics arguing that distributing readily available data to Chinese AI developers helps their models become as powerful as top U.S. models.
Ali Ansari, Micro1’s founder, stated last month on X that, unlike some of its competitors, the startup does not sell its data to Chinese model manufacturers. He commented, “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.”
Like Mercor, Micro1 initially started as an AI recruiting startup. However, after observing that data-labeling clients were using his AI platform to vet and recruit engineers for annotation, Ansari decided to pivot and also enter the data-labeling business.
Ansari previously informed TechCrunch that, in addition to having its experts evaluate model outputs — a concept known as reinforcement learning gyms — the company is constructing a robotics pre-training dataset by having hundreds of generalists record everyday object interactions in their homes.
Micro1 secured its Series A funding last September at a $500 million valuation, and TechCrunch understands that the startup may have recently completed another funding round at a significantly higher valuation.
Micro1 did not respond to a request for comment.