Google’s $1.5B+ Mechanize Deal: The Founder Succession Risk
Google’s reported $1.5 billion-plus Mechanize deal took its CEO and more than a dozen staff without buying the company outright. If that can happen to them, your succession plan is probably fiction.
Google’s reported $1.5 billion-plus Mechanize deal took its CEO and more than a dozen staff without buying the company outright. If that can happen to them, your succession plan is probably fiction.
Google has completed a talent deal with AI startup Mechanize that moved co-founder and former CEO Tamay Besiroglu into Google DeepMind as a research scientist. More than a dozen former Mechanize employees have also joined Google, largely working on mid-training efforts. Business Insider previously reported that Google had been discussing a transaction worth more than $1.5 billion for Mechanize’s technology and talent, though the final terms have not been disclosed. ([businessinsider.com](https://www.businessinsider.com/google-completes-deal-for-ai-agents-startup-mechanize-2026-9))
Mechanize’s former chief of staff, Guive Assadi, now lists himself as CEO. That is a leadership transition, technically. But let’s not kid ourselves: this is a brutal lesson in what happens when a giant decides your best people are more valuable inside its walls than competing against it.
Google bought the engine, not necessarily the car
Mechanize was not some sleepy software business that had run out of puff. The San Francisco startup launched in 2025 with an ambition to automate valuable work across the economy, beginning with software engineering. Earlier this year it said it had raised $9.1 million at a $500 million valuation, with backers including former GitHub CEO Nat Friedman, Stripe CEO Patrick Collison and podcaster Dwarkesh Patel. ([businessinsider.com](https://www.businessinsider.com/google-mechanize-deal-talent-tech-ai-coding-2026-8))
Then came Google.
Besiroglu is now at DeepMind. His former colleagues are there too. The original reporting said Google was discussing a non-exclusive licence for Mechanize technology alongside the hiring of its people. That distinction matters. A full acquisition has a clean headline: Company A buys Company B. A licence-and-talent transaction is messier. It can leave a company alive on paper while stripping out the people who created its technical edge.
That may be commercially sensible for everyone involved. I’m not saying it isn’t. Founders, employees and investors are allowed to take a big win when one lands on the table. A $1.5 billion-plus reported deal around a company that had recently raised $9.1 million is not something you dismiss because it offends your preferred founder mythology.
But calling it merely a win skips the operator’s question: what is left for the customers, the remaining staff and the new CEO to actually build?
That is Guive Assadi’s problem now. Not Besiroglu’s. Besiroglu has gone from CEO of a startup with a grand mission to research scientist at one of the world’s most powerful AI organisations. Assadi gets the title at the company left standing.
Titles are cheap. Control of the mission, product and A-team is expensive.
Why Sundar Pichai keeps making these AI talent deals
Google CEO Sundar Pichai is not buying talent for decoration. Google wants to improve at AI coding and agents, an area where Business Insider reported it has struggled relative to OpenAI’s Codex and Anthropic’s Claude Code. Mechanize’s work is useful because it helps improve how AI models perform on coding tasks. ([businessinsider.com](https://www.businessinsider.com/google-completes-deal-for-ai-agents-startup-mechanize-2026-9))
That means the prize is not simply a clever founder or a handful of frighteningly bright engineers. It is the machinery around them: evaluation systems, training workflows, hard-won judgment about where models fail, and the ability to make them less useless in real work.
Most people look at AI and obsess over the model. Fair enough — the demos are flashy. But models are becoming the visible layer of a much bigger industrial system. The advantage increasingly sits in proprietary data, evaluations, workflows, distribution and the people who know how to improve all four without setting money on fire.
Google has already shown the playbook. In 2025, it paid $2.4 billion to license Windsurf technology and hire its co-founder Varun Mohan and other senior researchers after OpenAI’s proposed $3 billion purchase of Windsurf fell apart. Mohan went on to lead Google’s agentic coding program, Antigravity. ([businessinsider.com](https://www.businessinsider.com/windsurf-google-cognition-acquisition-ai-coding-developer-data-ide-2025-7))
That is not a random shopping spree. It is a management strategy.
Big Tech can no longer assume it can buy whatever it wants in one neat acquisition. Regulators watch. Competitors complain. Lawyers earn their holiday houses. So the giants have found another route: licence the technology, hire the leadership, absorb the scarce expertise and let the rump company work out its own future.
From Google’s side, this is ruthlessly logical. It gets capability faster than building it internally, avoids some of the friction of a conventional acquisition and puts the people closest to the problem inside DeepMind.
From the startup side, it creates a question every founder should ask before the champagne appears: are you selling the business, or are you selling the bits that made it a business?
The overlooked angle: the CEO job just got harder
The easy take is that Besiroglu “exited” and Assadi “took over.” That makes for a tidy org chart. Real life is less tidy.
A new CEO inheriting a company after its founder and technical core move to a strategic partner does not inherit a normal scale-up. He inherits a credibility test. Customers will want to know whether the roadmap still exists. Employees will want to know whether there is a future worth staying for. Investors will want to know whether the remaining entity has a defensible asset, or merely a corporate shell and some legal documents.
This is where founders get too sentimental about succession.
A succession plan is not a press release naming a replacement. It is proof that the company can make decisions, win customers and retain talent when the original gravitational force disappears. If the business only works because one person is in every crucial meeting, knows every customer, makes every technical call and personally recruits every strong hire, you have not built a company. You have built a dependency with a cap table.
I’ve seen this mistake in businesses far less glamorous than AI. The founder tells everyone they are empowering the team, then quietly remains the only person who can price the deal, fix the product, calm the lender and make the big call. They are not indispensable because they are a genius. They are indispensable because they have failed to build an operating system without themselves at the centre.
Google did not create that weakness at Mechanize. But this sort of deal exposes it quickly.
Don’t romanticise the “acquihire on steroids”
There is another uncomfortable angle here. These arrangements may be good for elite staff and early investors, while being far murkier for everyone else.
Business Insider described the recent AI wave as a period in which large companies have used hybrid arrangements — talent hiring plus technology licensing — partly because standard acquisitions receive more antitrust scrutiny. Its reporting on Windsurf noted that the businesses left behind after similar deals can struggle, even though Windsurf itself found a buyer in Cognition for its remaining operations. ([businessinsider.com](https://www.businessinsider.com/windsurf-google-cognition-acquisition-ai-coding-developer-data-ide-2025-7))
That does not mean every deal is dodgy. It means founders should stop treating deal structure as a legal footnote.
Structure determines who owns the customers. It determines who keeps the intellectual property. It determines whether employee equity has a real outcome. It determines whether the remaining team has a proper business or a motivational poster saying “onward.”
The contrarian view is this: a founder’s duty is not to preserve independence at all costs. That is ego dressed up as principle. If Google offers a transaction that genuinely changes lives and gives the technology a better chance to matter, take it seriously.
But a founder also has a duty to negotiate the afterlife of the company with the same aggression used to negotiate the price.
Who leads the remaining business? What product rights does it retain? How are customers supported? What retention packages protect the people not selected by the buyer? What happens to the brand? Is there enough capital and enough talent left to fulfil the promises already made?
If you cannot answer those questions before you sign, you are not doing a deal. You are hoping.
What this means for you
If you run a company, do these four things this week.
First, run the “founder disappears” drill. Pick a Monday six months from now and assume you are gone. Who owns sales, product, hiring, finance and the three biggest relationships? Write the names down. If the answer to more than two is “me,” that is your real job list.
Second, identify your crown-jewel people and systems. Not the people with the loudest LinkedIn profiles — the people whose departure would make execution slower, riskier or dumber. Then work out why their knowledge is trapped in their heads. Document it, cross-train it and give them reasons to stay.
Third, treat strategic interest as a negotiation about continuity, not just price. If a bigger company wants your talent and technology, insist on a specific plan for the customers and staff left behind. Put it in the deal documents. Vague verbal reassurance is worth exactly what it costs.
Finally, build something that has a reason to exist without you. That is the real test. Not revenue. Not valuation. Not applause at a conference. A business is durable when a great operator can step in, understand the machine and make it better.
Google’s reported $1.5 billion-plus Mechanize deal is a reminder that in a hot market, your best outcome can also reveal your biggest weakness. Build for the payday if you like. But build the company so it survives one too.