Micro1 Went From $7 Million to $500 Million in Revenue in a Year

Micro1 has reached a $500 million gross annual run rate, up from $100 million late last year and $7 million at the start of 2025, according to TechCrunch. The AI data-labeling company, which recruits doctors, lawyers, and engineers to evaluate AI model responses, also made a $12.5 million offer for Spirit Airlines' corporate data in bankruptcy court, competing with Google's $10 million bid. Founder Ali Ansari initially built Micro1 around AI recruiting before pivoting to AI training work.

Micro1's $500 million gross run rate is a sharp signal from the AI market: labs still need paid human experts, even while they sell you the idea that models are learning on their own. Micro1 is no longer just another fast-growing AI vendor. Business Insider reported on August 21 that the company has made a $12.5 million offer for Spirit Airlines' corporate data, trying to beat Google's winning $10 million bid in bankruptcy court. That is a strange auction on its face. It also tells you exactly where this business is going. According to TechCrunch, Micro1 has reached a $500 million gross annual run rate, up from $100 million late last year. The company started 2025 at roughly $7 million in annual revenue, then crossed $50 million by September, when it raised $35 million in a Series A round led by 01 Advisors at a $500 million valuation. That's fast. Even by AI standards, it's fast. Here's what Micro1 actually sells. AI labs need humans to judge model answers before those models get trusted with harder work. Someone has to compare two chatbot responses, write the better answer to a coding problem, or spot when a medical explanation sounds fluent but goes wrong in the reasoning. That's the business. Micro1 recruits doctors, lawyers, engineers and other domain experts, then uses its AI recruiting tool, Zara, to screen workers before they handle training and evaluation tasks. The company didn't begin as a data-labeling specialist. Forbes reported in December that founder Ali Ansari first built Micro1 around AI recruiting, then pivoted after seeing customers use the system to find people for AI training work. That detail matters because it explains the company better than any grand claim about artificial intelligence. Micro1 found the bottleneck and moved toward it. Harvey Built Its Own Legal AI Model Instead of Renting One From OpenAI https://startupfortune.com/harvey-built-its-own-legal-ai-model-instead-of-renting-one-from-openai/ Harvey, the legal AI startup valued near $15.5 billion, has launched Tenet, its first proprietary model, built on a customized Kimi K3 base and post-trained on attorney-generated case files. The model ships inside a broader Harvey II relaunch that also introduces a Memory feature for law firms. - why legal AI startups build their own models https://startupfortune.com/harvey-built-its-own-legal-ai-model-instead-of-renting-one-from-openai/ - how law firms fine-tune AI with case files https://startupfortune.com/harvey-built-its-own-legal-ai-model-instead-of-renting-one-from-openai/ Gross revenue flatters the story So look past the headline. Gross run rate is not the same thing as software revenue, because these marketplaces pay a large share of customer money back to contractors. The Information has reported that rivals such as Mercor and Handshake pay roughly 60% to 70% of gross sales to the experts doing the work, leaving much smaller net revenue after contractor payments. That doesn't make Micro1 weak. It does mean you should not read a $500 million gross run rate like a $500 million SaaS business with fat margins and tiny delivery costs. Still, the surge is real. Sacra estimates Micro1 hit $300 million in annualized revenue in April 2026, up from $125 million at the end of 2025. TechCrunch reported last September that Micro1 was generating $50 million in annual recurring revenue, up from $7 million at the start of that year. Put those figures beside the latest $500 million number and you get a company that has been repriced almost every quarter by demand from AI labs. The timing helped. Meta's roughly $14.3 billion investment in Scale AI in June 2025 unsettled a market where the biggest AI labs were already nervous about sending sensitive training work to a rival's partner. Bloomberg reported that OpenAI was phasing out work with Scale after the Meta deal, while TechCrunch later reported that OpenAI and Google said they would stop working with the data provider. Scale has said it remains independent and will not give Meta access to customer confidential information. Rival labs were never going to take that entirely on faith. The market is bigger than Micro1 Micro1 is not winning. Mercor founder Brendan Foody said in July that Mercor had crossed $2 billion in annualized revenue run rate, and The Information reported that Handshake's AI-training business had reached nearly $1 billion in gross annualized revenue earlier this year. That's a crowded field. It also makes Micro1's rise more interesting, not less, because several companies are growing at once from the same pressure point. The pressure point is human judgment. Frontier models can generate answers at scale, but labs still need people who know when an answer is subtly wrong. A senior lawyer checking contract analysis is not doing the same job as a casual tester clicking a thumbs-up button. A physician reviewing a clinical answer is carrying a different kind of liability. Labs pay for that gap because benchmarks don't catch everything and customer trust doesn't survive many embarrassing failures. The Spirit Airlines bid gives the story a harder edge. Business Insider reported that Micro1 is trying to persuade the bankruptcy court to consider its higher offer even though it missed the initial bid deadline, with a new hearing set for September 9 after privacy concerns delayed the auction. If the court entertains the offer, Micro1 would not just be supplying experts to clean up model behavior. It would be buying messy corporate data that could help train AI agents on real operational problems. That is the bet. The next edge in AI may come less from another model demo and more from controlling the data, the expert labor and the evaluation environments that make models useful. GPUs still matter, of course. But if you want to understand why Micro1 can grow this quickly, don't start with the machines. Start with the people paid to tell the machines when they're wrong. Accounting startup Rillet hits $1 billion valuation as AI eats CFO busywork https://startupfortune.com/accounting-startup-rillet-hits-1-billion-valuation-as-ai-eats-cfo-busywork/ Rillet, an AI-native accounting platform founded by former N26 U.S. CEO Nicolas Kopp, raised $100 million at a $1 billion valuation in a Series C led by ICONIQ. The startup automates month-end close and bookkeeping for finance teams, with customers including Mercor, Neuralink and Skild AI. - how to automate accounting tasks with AI software https://startupfortune.com/accounting-startup-rillet-hits-1-billion-valuation-as-ai-eats-cfo-busywork/ - AI accounting platform raises 1 billion dollar valuation https://startupfortune.com/accounting-startup-rillet-hits-1-billion-valuation-as-ai-eats-cfo-busywork/ Also read: ChatGPT Can Now Read and Send Your iMessages on a Mac https://startupfortune.com/chatgpt-can-now-read-and-send-your-imessages-on-a-mac/ • AI Data Centers Become a $130 Billion Political Liability in 2026 https://startupfortune.com/ai-data-centers-become-a-130-billion-political-liability-in-2026/ • Ramp Launches a Free AI Model Router That Already Cut Client Costs 40% https://startupfortune.com/ramp-launches-a-free-ai-model-router-that-already-cut-client-costs-40/