Micro1's rapid rise over the past two years has placed it amongst a cohort of AI corporations which are scaling at breakneck speed. The three-year-old startup, which helps AI labs recruit and manage human experts for training data, began the 12 months with annual recurring revenue (ARR) of around $7 million.
Today, ARR is claimed to have exceeded $100 million, founder and CEO Ali Ansari told TechCrunch. That figure can be greater than double the revenue Micro1 reported in September when it announced its $35 million Series A at a $500 million valuation.
Ansari, 24, said on the time that Micro1 was working with leading AI labs, including Microsoft, in addition to Fortune 100 corporations struggling to enhance large language models through post-training and reinforcement learning. Their demand for world-class human data has fueled a fast-growing market that Ansari predicts will grow from $10 billion to $15 billion today to almost $100 billion inside two years.
The rise of Micro1 and bigger competitors like Mercor and Surge accelerated after OpenAI and Google DeepMind allegedly lost connection to Scale AI following Meta's $14 billion investment within the provider and the choice to rent Scale's CEO.
According to the founder, Micro1's ARR is growing quickly, but has not yet caught up with its competitors: Mercor, with greater than $450 million, sources told TechCrunch, and Surge reported $1.2 billion in 2024.
Ansari attributes Micro1's growth to its ability to quickly recruit and evaluate subject material experts. Like Mercor, Micro1 began as an AI recruiter called Zara that matched technical talent with software roles before entering the information training market. This tool now interviews and screens applicants in search of expert positions on the platform.
In addition to providing expert-level data to leading AI labs, Ansari says two recent segments, barely visible today, are poised to reshape the human data economy.
Techcrunch event
San Francisco
|
thirteenth–fifteenth October 2026
The first will likely be non-AI native Fortune 1000 corporations that can begin developing AI agents for internal workflows, support operations, finance, and industry-specific tasks.
The development of those agents requires systematic evaluation: testing frontier models, evaluating their results, choosing winners, fine-tuning these agents, and continually validating performance in production. Ansari argues that this cycle relies heavily on human experts assessing AI behavior at scale.
The second is robotics pre-training, which requires high-quality, human-made demonstrations of on a regular basis physical tasks. Micro1 is already constructing what Ansari calls the world's largest robotics pre-training dataset, collecting demonstrations from a whole lot of generalists who record object interactions of their homes. Robotics corporations will need large amounts of this data before their systems can function reliably in homes and offices, he said.
“We expect a big portion of product budgets in non-AI-native corporations to go toward analytics and human data, from 0% to no less than 25% of product budgets,” said the CEO, who founded Micro1 while at UC Berkeley. “We also support robotics labs in creating robotics data; these two areas will represent an enormous share of this $100 billion per 12 months market.”
Even as recent markets emerge, Micro1's current growth still comes primarily from elite AI labs and AI-focused corporations. The startup is expanding its work with these labs on reinforcement learning, the feedback loop for testing and improving model behavior.
Micro1 hopes its early entry into robotics data and enterprise agent development, in addition to the scaling of its specialized RL environments, will help it capture additional market share as the information wars heat up.
For now, Ansari says the corporate is concentrated on scaling responsibly, paying experts well and keeping people at the center of an industry built on training machines.
The company currently manages 1000’s of experts in a whole lot of fields, starting from highly technical areas to surprisingly offline disciplines. Many earn nearly $100 an hour, in accordance with Ansari.
“There are Harvard professors and Stanford graduate students who spend half their week training AI through Micro1,” Ansari said. “But the larger change is within the sheer scope and range of roles. It's expanding into areas you wouldn't expect to be relevant to language model training, including offline and fewer technical areas. We're very optimistic concerning the development.”
