Data‑labeling firms Handshake and Mercor near $1B run‑rates as AI demand soars

Rapid revenue surge, according to reports

It has been reported that data‑labeling startup Handshake’s gross annualized revenue recently reached roughly $1 billion — up from about $550 million in January. It has also been reported that Mercor hit a $1B+ gross annualized revenue pace this year. These figures, disclosed to The Information and picked up on Techmeme, point to blistering growth in a corner of the AI supply chain that rarely makes headlines.

What’s driving the bump?

Simple answer: demand. Companies building and fine‑tuning large models need huge volumes of labeled, high‑quality human‑annotated data, and — for now — humans still do that job best. Contracts for human contractors, rapid scaling of labeling workforces, and enterprise clients hungry for improved model performance are all pushing revenue through the roof. It’s the AI gold rush, but instead of pickaxes it’s keyboards and annotation tools.

Risks, costs and a human story

Fast growth is great on paper. But can margins hold? Will worker supply, quality control and data‑privacy risks keep pace? Allegedly, many of these firms are racing to scale their contractor networks and automation tools simultaneously — a tricky balancing act. There’s a social angle, too: the invisible army of labelers powering today’s chatbots and image models are suddenly central to a multibillion‑dollar industry. That feels important. And uneasy.

Why this matters

If these run‑rate figures are real, they’re a clear signal: the market for human‑in‑the‑loop services is no niche. It’s a backbone for much of modern AI. So here’s the question: as models get smarter, who will build the scaffolding — and who will benefit when the next wave of automation arrives? The stakes are higher than a quarterly headline. This is about where the value in AI actually lands.

Sources: theinformation.com