Handshake Hits $1B Gross Revenue

Handshake hitting the same gross revenue milestone as Mercor shows that expert data labeling has become a scale business, not a niche add on. Handshake used a student and postdoc network built for campus recruiting, then redirected it into frontier lab work where customers need math, law, and science experts to grade model outputs, write better answers, and check reasoning chains. By April 2026, that business had overtaken its legacy recruiting segment and pushed total gross annualized revenue to about $1.1B.1 2 3 4

• This is not the same as $1.1B of software revenue. In this market, reported topline usually includes contractor payouts. Handshake was estimated at about $450M of net revenue after paying workers, which matters more for durability because customer switching costs are low and gross margins on labeling are much lower than on software.1 4
• Handshake and Mercor arrived here from different supply engines. Mercor was built around matching pre vetted experts to AI lab tasks. Handshake started with 18M students and alumni across 1,400 colleges, then tapped PhDs, graduates, and postdocs already inside that network. That made Handshake look less like a new marketplace and more like a repurposed expert inventory with immediate depth.1 2 5
• The next fight is over quality infrastructure, not just worker count. Handshake bought Cleanlab to move from supplying people to checking whether labels are correct and building evaluation workflows. That mirrors where the market is going, as buyers increasingly want human data that is auditable, fast, and embedded into model testing rather than one off annotation jobs.1 3 4

Going forward, the winners in human data will look less like staffing firms and more like workflow software layered on top of rare expert supply. Handshake now has the scale to compete for the biggest lab budgets, and its path to staying there is to turn one time labeling spend into a repeat system for evaluations, safety checks, and ongoing model improvement.1 3 4