Handshake, a 12-year-old job platform, has seen AI training revenue surge to nearly $1 billion annually, up from $550 million in January, as AI companies demand human contractors to grade model outputs.
The explosive growth underscores AI firms' insatiable need for quality training data. Handshake now contracts with lawyers, PhD holders, medical doctors and other specialists to evaluate AI-generated answers—work essential for improving large language models and other AI systems.
The startup, which originally served as a job site for college graduates, entered AI training just over a year ago when revenue stood at $5-10 million annually. The 80-fold increase reflects AI's rapid scaling demands and the shortage of available talent for this labor-intensive work.
Handshake is not alone. Mercor, another contractor platform, is similarly capitalizing on the trend. These startups fill a critical gap between AI developers needing human feedback loops and specialized professionals seeking flexible contract work.
The model reveals a hidden layer of AI infrastructure: while headline companies build models, less visible firms quietly scale the human workforce powering their development. As AI competition intensifies, demand for quality training data shows no signs of cooling.
Amsterdam-based Monumental secured a $32 million Series B round led by Khosla Ventures. The company develops autonomous robotics and AI software for the construction industry.
Indian AI coding startup Emergent reached a $1.5 billion valuation in its latest funding round, raising $130 million in Series C and joining the unicorn club after a five-fold valuation jump in six months.
Rime, an AI platform processing over 100 million calls monthly, secured $24 million in Series A funding to expand its customer service solutions for enterprises.
A former SpaceX engineer has secured $65 million in funding to revolutionize wire harness manufacturing for aerospace and defense. The startup aims to replace decades-old production methods still used in rockets, missiles, and satellites.