Handshake AI Reaches Nearly $1B ARR in One Year
Garrett Lord and the Handshake AI team discussed how they built an AI training data business generating nearly $1 billion in gross annualized revenue within roughly a year. The deep dive explores what frontier labs are currently purchasing specifically for reinforcement learning. This provides a rare and high-value financial and technical look into the booming AI training data market at a time when traditional internet scraping is hitting bottlenecks. Understanding what top labs buy sheds light on the commercial drivers behind next-generation AI models. Handshake AI leverages academic depth by tapping into university talent pipelines to supply graduate-level reasoning data in hard sciences. They also acquired Cleanlab to integrate automated label auditing directly on top of human contributor workflows.
## BACKGROUND
Frontier AI labs face a massive data bottleneck because publicly available internet data has been largely exhausted, making specialized human-generated data extremely valuable. Reinforcement learning, one of the core machine learning paradigms, requires high-quality interactive data and specialized rewards to train advanced reasoning agents effectively.