micro1 tops $1 billion annual run rate and prices Series B at $4 billion valuation

micro1 said it passed $1 billion in annual run rate while staying profitable and priced a Series B at a $4 billion valuation, with a tender offer for early employees. The San Francisco data lab said the milestone follows a near-collapse three years ago and comes as it expands its global team and doubles down on long-term AI data infrastructure.

Why it matters: - micro1 says it has crossed $1 billion in annual run rate while remaining profitable, a rare combination in AI infrastructure. - The Series B valuation of $4 billion signals continued investor appetite for companies building data, training and evaluation systems for frontier AI models. - The approximately $25 million tender offer gives early employees a chance to sell shares and lock in gains.

What happened: - micro1 announced on Oct. 12, 2026 that it surpassed $1 billion in annual run rate. - The company also said it priced its Series B after passing $500 million in annual run rate this summer. - micro1 operates from San Francisco with a fully remote core team of about 180 people across more than 14 countries. - Dream Ventures led the round as one of micro1’s earliest investors. - Microsoft, Javelin, Flybridge, G Squared, Up Partners and LG also participated.

The details: - micro1 describes itself as a data lab that trains frontier AI models and evaluates agents. - The company said it lost 60% of its revenue overnight about three years ago and nearly went bankrupt. - micro1 now says it has maintained close to 100% core team retention. - The company said it will slow headcount growth significantly even after reaching $1 billion in annual run rate. - micro1 tied its growth to an operating model built around long-term investment in research, products and infrastructure. - The company laid out eight operating principles in its Q3 2026 update, including human-first AI, U.S.-first data rules, long-term research over short-term revenue, and stronger incentives for exceptional effort. - micro1 also said it wants more abundant and affordable training data, plus more realistic model training environments and longer model horizons. - The company said it believes data could become the largest industry in history.

Between the lines: - The funding round and valuation suggest investors see durable demand not just for model training, but for the picks-and-shovels layer that supplies data, evaluation and reinforcement learning environments. - micro1 is signaling discipline rather than hypergrowth: a lean team, a streamlined board and slower hiring all point to a company trying to protect margins and execution. - The U.S.-first rule and refusal to work with adversarial nations place micro1 in the center of the broader AI supply-chain and national security debate.

What's next: - micro1 said it will keep investing in people, research and infrastructure to support its long-term data strategy. - The company plans to slow team growth while continuing to expand data generation capabilities and training environments. - micro1’s next test is whether it can sustain rapid revenue growth and profitability while staying small relative to its scale.

The bottom line: - micro1 is using a major valuation step-up to underscore a simple bet: the infrastructure behind AI models may be as valuable as the models themselves.