As AI labs abandon Scale AI over Meta ties, Micro1 emerges as a high-growth competitor offering expert-sourced training data and new AI simulation environments.
Micro1 Secures $35M to Scale Expert-Led AI Data Labeling
Micro1, a fast-growing startup helping AI labs find and manage human contractors for data labeling and training, has raised a $35 million Series A round, pushing its valuation to $500 million.
The round was led by 01 Advisors, the venture firm co-founded by former Twitter executives Dick Costolo and Adam Bain, the latter of whom is also joining Micro1’s board. Joshua Browder, founder of DoNotPay, will also take a board seat.
This latest funding underscores Micro1’s role as a rising alternative to Scale AI, especially as AI labs distance themselves from Meta’s influence.
The Scale AI Fallout Creates Space for New Leaders
Micro1’s rise comes at a moment of industry realignment. After Meta invested $14 billion in Scale AI and brought its CEO on board, leading AI labs such as OpenAI and Google reportedly began cutting ties with the data provider due to concerns over data confidentiality and research leakage.
While Scale AI denies any impropriety, the perception of risk has opened the door for startups like Micro1, Mercor, and Surge to capture market share.
“Really the only way models are now learning is through net new human data,” said Adam Bain.
“Micro1 is at the core of providing that data to all frontier labs.”
Strong Growth and High-Quality Focus
Led by 24-year-old CEO Ali Ansari, Micro1 is already working with Microsoft and several Fortune 100 clients, and has scaled from $7M to $50M in annual recurring revenue (ARR) in less than a year.
Although smaller than competitors like:
- Mercor ($450M+ ARR)
- Surge ($1.2B ARR in 2024)
…Micro1 is growing fast, and carving out a niche by prioritizing high-skill, domain-specific labeling over the low-cost, high-volume approach Scale AI pioneered.
Ansari argues the data needs of AI labs have evolved. Today’s foundation models require:
- Expert-labeled data (from software engineers, doctors, writers)
- Quality over quantity
- Domain expertise for complex tasks
Zara: Micro1’s AI-Powered Recruiting Engine
To meet this demand, Micro1 built an AI recruiter named Zara, which interviews, vets, and matches experts with appropriate AI training projects.
Zara has already recruited thousands of qualified contractors, including:
- Professors from Stanford and Harvard
- Specialists in software, law, medicine, and more
The company plans to add hundreds more experts each week, helping labs fine-tune large models with better-aligned, more accurate data.
Next Frontier: AI Training in Virtual Environments
As the AI training market matures again, Micro1 is preparing for what’s next: interactive training environments. These virtual workspaces allow AI agents to learn by completing simulated tasks — essential for developing next-generation, agentic AI systems.
Ansari says Micro1 is actively building products for this emerging vertical, where training involves doing, not just seeing or reading.
This move aligns with a broader shift in AI development from static, text-based datasets to dynamic, task-driven learning environments.
A Fragmented Market with Room to Grow
One advantage for Micro1 and its peers? No single provider can satisfy the entire data appetite of frontier AI labs. Companies like OpenAI, Anthropic, and Google often work with multiple training data vendors, creating a non-zero-sum market — for now.
That gives Micro1 room to grow without needing to unseat incumbents entirely.
“The nature of the business is such that it’s difficult for any one company to handle all of one AI lab’s data needs,” said Ansari.
What’s Next for Micro1
With fresh funding and a surge in demand, Micro1 plans to:
- Expand its expert contractor pool
- Invest in virtual environment training tools
- Deepen relationships with top AI labs and enterprises
- Differentiate through data quality, speed, and trustworthiness
As frontier models grow in complexity and importance, Micro1’s position as a trusted, flexible, and expert-driven data provider may only become more valuable.
