On Tuesday, Jeff Dean — the closest thing Google had to a living legend in its engineering ranks — posted on X that he was leaving the company after 27 years. He is co-founding Discovery Loop, a public benefit corporation, alongside longtime collaborators Sanjay Ghemawat, Oriol Vinyals, and Quoc Le. The mission: “automate machine learning, science, and engineering to accelerate discoveries and progress.” Alphabet’s stock dropped roughly 5%, erasing somewhere between $160 and $200 billion in market value before the closing bell.
The headlines wrote themselves. “Brain Drain at Google.” “AI’s Biggest Name Goes Rogue.” “The End of the Search Giant’s AI Dominance.”
None of them mentioned the most important detail buried in the announcement: Google is a founding investor. Google Cloud is the compute provider. This is not a departure. It is a spin-out with better branding.
The PBC as Corporate Escape Hatch
Discovery Loop is structured as a public benefit corporation — a legal form that ostensibly obligates the company to pursue a social mission alongside shareholder returns. It is a structure associated with do-gooder startups, B Corps, and mission-driven founders who want to signal that profit is not their only motive.
It is also, increasingly, the preferred vehicle for big tech companies that want to shed liability without shedding upside.
Here is how it works. A team of elite researchers inside a company like Google wants to build something ambitious — something that might take a decade, might fail spectacularly, and might, in the process, produce technology that regulators, plaintiffs’ attorneys, and congressional committees would love to depose the parent company about. The parent company cannot kill the project without losing the team. It cannot keep the project inside without accepting the legal and reputational exposure. So it spins the team out into a PBC, writes a check, locks in a cloud contract, and retains a stake. The four founders get equity and autonomy. The parent gets distance and optionality. Everyone wins — except, perhaps, the public that the “public benefit” label was supposed to serve.
One Bay Area attorney who has structured several such deals in the past two years put it to me this way, standing outside a Palo Alto courthouse between depositions: “The PBC is the new dual-class share structure. It lets you say you’re doing good while making sure nobody can force you to actually do it.”
A $180 Billion Overreaction, or a Perfectly Rational One?
The market’s reaction was swift and brutal. A single personnel move — not an earnings miss, not a product failure, not a regulatory action — vaporized nearly $200 billion in shareholder value in a matter of hours. If that number seems absurd, consider what it implies about how the market actually values Alphabet.
It is not valuing the data centers. It is not valuing the search monopoly, the YouTube ad inventory, or the Android install base. It is valuing, with startling precision, the belief that a handful of people — perhaps a few dozen, perhaps fewer — are the only thing standing between Google and irrelevance in the AI race. When four of them walk out the door together, the market does not shrug. It panics.
This is not a vote of confidence in the corporate form. It is an admission that the corporate form, for all its quarterly earnings calls and investor relations polish, is a thin membrane stretched over a handful of irreplaceable individuals. The moat is not the infrastructure. The moat is the people. And people, unlike server racks, can walk.
The Spin-Out Is the New Golden Parachute
There was a time when the ultimate exit for a big-tech executive was a retirement announcement, a generous severance package, and a board seat or two. That model is starting to look quaint. The new playbook — the one Dean and his co-founders just executed — is cleaner, more lucrative, and far more interesting.
You do not retire. You do not join a competitor. You let your employer invest in you, take their cloud contract, and set up shop next door with a mission statement that sounds like it was drafted by an AGI safety nonprofit. You get founder equity in a vehicle that can raise capital at a valuation untethered from the parent company’s P/E ratio. The parent gets to tell Wall Street it still has exposure to your work while telling regulators it has no control over what you build. The PBC wrapper provides a rhetorical shield against accusations of pure profit-seeking, even as the economic structure looks an awful lot like a corporate R&D lab with a better LinkedIn presence.
Discovery Loop’s first non-Google-led funding round will be the real test. If it prices at a valuation north of $5 billion — a number already being floated in private conversations, according to Reuters — then the spin-out model will have proven itself not just as a retention tool but as a wealth-creation engine that makes the old IPO path look sluggish.
The uncomfortable question nobody is asking: if this structure is so good for everyone involved, what exactly was the “public benefit” that required a separate legal entity? What can Discovery Loop do as a PBC that Google Brain could not do as a division of Alphabet? The answer, almost certainly, is not about mission. It is about liability, compensation, and the freedom to build things that a publicly traded company with a $2 trillion market cap would rather not have on its own balance sheet.
Jeff Dean did not escape Google. Google let him leave — and made sure it kept a key to the door. The rest of us are just watching from outside, trying to figure out who is really in the driver’s seat.