Fetching from the wire…
Public story · 2026-09-12 · high
Hiring the team and licensing the tech let Google skip the merger filing a straight acquisition would have required.
Why now: Business Insider reported the deal as closed, with Mechanize's cofounder already reassigned inside DeepMind.
Google closed a deal worth more than $1.5 billion for Mechanize, an AI coding-agent startup, without filing anything with regulators. The company hired more than a dozen Mechanize staff and took a non-exclusive license to its technology instead of acquiring it outright. That structure sidesteps merger review entirely, per Business Insider.
Mechanize's cofounder and former CEO Tamay Besiroglu is now a research scientist at DeepMind, working on midtraining. Former chief of staff Guive Assadi lists himself as CEO of whatever remains. Besiroglu founded Mechanize in 2025 with Matthew Barnett and Ege Erdil to build environments, benchmarks, and training data for long-horizon agents. The startup raised $9 million in a seed round. It reached this deal in about a year.
License-plus-hire is now the standard shape for a big lab to absorb a team without a merger filing. This deal is why it keeps working. Business Insider's report doesn't say whether cofounders Matthew Barnett and Ege Erdil joined Google too, or if they're sitting this one out.
Each link below shares sources, entities, or timing with this story.
The former Google chief scientist is raising at roughly $50 billion, up from about $10 billion on a planned $1 billion raise only weeks ago. Cofounded with Sanjay Ghemawat, Quoc Le and Oriol Vinyals, the company pitches automated systems running massive parallel scientific exp...
The pilot puts evaluation inside Confidential Space on Google Cloud's Confidential Computing so Gemini Flash Lite's weights stay private from evaluators while evaluator prompts stay private from Google (DeepMind). Partners are the Singapore AI Safety Institute, OpenMined, AVER...
Six clients. One manifest. Zero vendor lock. Vercel published Agent Plugins 1.0.0 on August 6, an openly licensed spec that bundles Agent Skills and MCP servers behind a single portable manifest. The shape is deliberately boring: a plugin.json requiring only schemaVersion and...
DeepMind's commitment is denominated in tokens and cloud credits rather than cash, an in-kind structure that keeps the spend on Google infrastructure (DeepMind). OpenAI published its own commitment with the Department of Energy and the national labs the same week (OpenAI). Fed...
Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber, the last purpose-built to find and patch vulnerabilities and pitched as a cheap alternative to large security-specialized models like Mythos (DeepMind). Splitting a cheap tier into a security SKU is new packa...
At Google I/O 2026, Pichai admitted Google trails on agentic coding with tool use and long-horizon tasks, blaming the missing data-flow "surface area" that Anthropic got with Claude Code and Cursor. He framed Gemini 3.5 Flash, purpose-built for agents, as the catch-up. A front...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.