Google may be considering a deal worth more than $1.5 billion to bring in technology and talent from a startup with one audacious goal: teaching AI to do the work of software engineers.
Business Insider reported on Aug. 5 that Google was in advanced talks for a deal involving AI startup Mechanize that would give Google access to its technology and bring some of its employees into the company. The reported agreement would deepen Google’s investment in AI coding as competition with Anthropic, OpenAI, Meta, and other labs shifts from chatbots toward agents capable of completing real software work.
And this would not be Google’s first billion-dollar swing at the problem.
Google could strike a $1.5 billion-plus Mechanize deal
Business Insider reported that Google has discussed a deal worth more than $1.5 billion that would include a licensing agreement for Mechanize technology and the hiring of some of the startup’s employees.
Google and Mechanize declined to comment to Business Insider. The negotiations were still in progress when the report was published, and the publication cautioned that details could change. No completed agreement has been publicly announced.
The structure would also look familiar.
In July 2025, Google agreed to pay roughly $2.4 billion to license technology from AI coding startup Windsurf while bringing CEO Varun Mohan, co-founder Douglas Chen, and several researchers into Google DeepMind. Reuters reported at the time that the incoming team would work primarily on agentic coding initiatives involving Gemini.
A Mechanize agreement would therefore represent another major investment in outside technology and talent as Google tries to strengthen its position in one of AI’s fastest-moving markets.
Google has also been building internally. At Google I/O earlier this year, the company unveiled Gemini 3.5 Flash and new agentic coding tools aimed at competing with Anthropic and OpenAI. More recently, it released Gemini 3.7 Flash, with coding and AI-agent workloads as central selling points.
The reported Mechanize talks suggest that Google sees an opportunity to further accelerate those efforts.
Mechanize wants to automate software engineering
Mechanize is a particularly intriguing target because its ambitions extend well beyond those of another coding assistant.
Founded in 2025 by Tamay Besiroglu, Matthew Barnett, and Ege Erdil, the company develops virtual environments, benchmarks, and training data designed to teach AI agents to perform increasingly complex work.
Its ultimate mission is unusually sweeping. In Mechanize’s original announcement, the founders said they want to enable the “full automation of the economy.”
Software engineering is an early focus.
Mechanize argues that today’s coding agents can perform impressively on narrow tasks but still struggle with the long-running, messy work humans encounter inside real software projects. In its explanation of its software-engineering strategy, the company points to challenges including reliability, long-context performance, and the ability to handle broader agentic tasks.
That mission has already put Mechanize on eWeek’s radar. When the startup launched, eWeek examined its controversial goal of eventually automating human work across the economy.
Mechanize announced in April that it had raised $9.1 million at a $500 million post-money valuation. A proposed deal valued at more than $1.5 billion would put a striking price on access to technology and talent from a company that is barely more than a year old.
AI coding is becoming a Big Tech arms race
The bigger story is the value Big Tech is beginning to place on AI coding.
Anthropic has turned Claude Code into one of its most visible products, with capabilities that extend well beyond autocomplete to tasks such as navigating repositories, editing files, running tests, and coordinating agents. eWeek has previously broken down how Claude Code is changing the role of AI inside software development.
OpenAI is pushing Codex in the same direction, while Meta entered the competition this month with Muse Code, its dedicated AI coding agent.
The technology is also beginning to change how engineers at AI companies describe their own jobs. Engineers at Anthropic and OpenAI have said that AI now generates all or nearly all of the code they personally write, an extreme example of where vendors hope these tools are heading.
That makes software development more than another feature category for frontier models. It is becoming a proving ground for whether AI agents can perform valuable work autonomously enough that companies will pay for the output.
For Google, another billion-dollar-plus agreement would make the stakes unusually visible. The company already has Gemini, developer tools, massive cloud distribution, and the Windsurf talent it brought into DeepMind. If the reported Mechanize negotiations lead to a deal, Google would add another piece to that stack.
What this means for developers and tech leaders
For developers, the immediate takeaway is that AI coding tools are moving from optional assistants toward a more central role in software development.
As Google, Anthropic, OpenAI, and Meta invest more heavily in agents that can navigate codebases, make changes, run tests, and handle longer tasks, engineering teams will need to decide where these systems fit into their workflows and where human review still matters most.
For tech leaders, the shift raises a different set of questions. AI coding tools may help teams move faster, but they also raise new questions about tool standardization, security, code quality, vendor dependence, and the level of autonomy organizations are willing to give software agents.
Google’s reported Mechanize talks are therefore more than another AI acquisition-style maneuver. They are another sign that the race is shifting toward systems that can handle larger chunks of real engineering work, and companies may soon have to decide not only whether to use AI coding agents, but also how much of the development process they are comfortable handing over.
Also read: Google’s New Gemini Flash Model Targets the Cost of Building AI Agents to see how Google is using Gemini 3.7 Flash to compete on coding performance, agent capabilities, and price.