Google is in talks to snap up key talent and licensing rights from Mechanize, a two-year-old reinforcement learning startup, in a deal pegged at $1.5 billion to $2 billion. The move signals how hard Big Tech is pushing for the people and tools behind the next wave of autonomous coding agents.
Mechanize builds digital training grounds where AI models learn to write, debug, and ship code without human intervention. Think of it as a gym for frontier coding agents. The startup, founded in 2025 by Tamay Besiroglu, Matthew Barnett, and Ege Erdil, has already drawn backing from Google Chief Scientist Jeff Dean and Stripe CEO Patrick Collison.
The acqui-hire plus structure
Google isn't buying Mechanize outright. Instead, negotiations center on acquiring core talent while licensing the startup's underlying technology. This "acqui-hire plus" model gives the buyer both the engineers and the blueprints without absorbing the whole company. The approach may reflect regulatory caution, as antitrust pressure on Big Tech's AI deals continues to mount.
What makes Mechanize valuable goes beyond just coding. The company's stated ambition is full automation of work across industries, with software engineering as the launchpad. An AI system that handles feature building, testing, debugging, and deployment end-to-end without human oversight represents a fundamental shift in how code gets made.
This article is for informational purposes only and should not be construed as financial or investment advice.
