Are we watching the emergence of a completely new model for achieving category leadership in technology?
For decades, Big Tech’s instinct was straightforward – find the world’s best talent and acquire it. If necessary, give it extraordinary resources, and then work very hard to keep those people inside the organisation.
But what happens when you simply can’t keep them forever? And more importantly, what happens when trying to keep them actually becomes the wrong strategy. That’s what makes the current Demis Hassabis story so interesting.
A different approach to talent management
The question isn’t retention. It’s relevance
Hassabis isn’t really leaving Google DeepMind.
He’s stepping away from the day-to-day CEO role and moving into chairman and chief scientist positions at Alphabet, while continuing to lead Isomorphic Labs.
The important point is not the job title. It’s the philosophy behind the move. Rather than saying, “You’ve built this thing, so now you’re either the CEO or you’re out”, Google is effectively asking “What’s the next challenge someone like Demis Hassabis should be working on?”
That’s a very different approach to talent management. And potentially, a very different approach to category creation.
After all, DeepMind itself was essentially the original example of this model. Hassabis founded DeepMind, built it into one of the world’s most important AI research organisations, and Google acquired it.
The traditional acquisition playbook would be familiar – bring in the founder, integrate the company, give them increasing responsibility and, eventually, make them part of the corporate machine. But what happens when the person who built the machine is no longer best used running it? That is the critical distinction.
The question isn’t necessarily “How do we retain Demis Hassabis as an employee?” It’s “How do we keep Demis Hassabis doing something strategically important to us?”
Those are two completely different questions.
From employees to ecosystems
And that’s where the category-leadership argument gets really interesting.
If you’re Google, you don’t necessarily need every great scientist to spend the next 20 years climbing the corporate hierarchy. You need to remain the place where the world’s best people can take on the world’s hardest problems. Even if their relationship with the company changes.
Jeff Dean takes that idea even further. After nearly three decades at Google, Dean is leaving with a group of senior researchers to build Discovery Loop – with Google investing in the new company.
That is fascinating because, under the old model, a departure like this would automatically be interpreted as “We’ve lost Jeff Dean”. Under the emerging model, the question becomes “Can Google help Jeff Dean build the next important thing?”
That’s a profound shift.
The old Big Tech mindset was almost “If you leave, you’re taking intellectual capital with you.”
The emerging mindset might be “If you’re going to leave, let’s make sure you’re leaving to do something ambitious – and let’s find a way for us to participate.”
That creates the possibility of a much more powerful ecosystem.
Google can provide capital, infrastructure, compute, talent networks and distribution. The founder or researcher gets autonomy. And the new company can move faster than it might inside a giant organisation.
Call it a new kind of `close spin-off` architecture.
But is this really a new model?
There’s an obvious counterargument. Perhaps this is simply a clever way of putting a positive spin on a talent exodus.
Google is losing extraordinarily important AI people. Alongside Dean, there are figures such as Noam Shazeer and John Jumper, while OpenAI and Anthropic continue to move at extraordinary speed.
So perhaps this isn’t a revolutionary new model for category leadership. Perhaps it’s a company with a weakened core trying to make the best of losing people.
And I think both things can be true.
That’s the Schrödinger’s-cat approach to category creation – the organisation can simultaneously be experiencing a serious talent-retention problem and discovering that trying to lock its best people into one organisational structure may be counterproductive.
Look at Hassabis.
If his comparative advantage is thinking about AGI, science and the long-term direction of AI, then having him spend every day managing an enormous organisation may simply not be the highest-value use of his time. And, perhaps, it’s not what makes him happiest either. That distinction matters far beyond Google.
The founder doesn’t have to run the machine
One of the biggest mistakes companies make is assuming that the person who created the category should also run the organisation forever. But those are different skills.
The person who sees the future and creates the opportunity isn’t necessarily the person who should run the operating machine that delivers it. That’s where Koray Kavukcuoglu, DeepMind’s Chief AI Architect becomes important.
Operational leadership moves to someone deeply embedded in DeepMind’s technology organisation.
In effect, two jobs are being separated; one person looks further out while another runs the engine.
It sounds obvious. But it’s actually quite radical in founder-led technology companies, where leadership is often equated with control and the CEO is the person with the most control. bPerhaps the next generation of category leaders will be less interested in control and more interested in optionality.
The company as a network
I love this idea because the best technology companies increasingly look less like pyramids to be climbed and more like networks to be maximised. The goal becomes having the world’s best people connected to your ecosystem.
They don’t necessarily all have to report to you. They don’t necessarily all have to be employees. They might be founders, investors, researchers, spin-outs, partners or suppliers. The strategic asset becomes the network itself.
And AI accelerates this shift. The half-life of an organisational structure is getting shorter.
You can spend five years building a world-class AI team, only to discover that the frontier – and your ambition – has moved somewhere else.
One researcher wants to work on foundation models. Another wants to work on biology. Another wants to automate software engineering. Another wants to build a company.
If the parent organisation says,”No, you’re staying here because we need you,” eventually the talent walks. But what if the response is “Okay. What’s the biggest problem you want to solve next — and how can we help?”
You may actually preserve much more of the relationship. Not necessarily the employee. The relationship. And I suspect that distinction is going to become incredibly important.
The new definition of category leadership
So perhaps this is the new definition of category leadership. Not an obsession with owning the category. But becoming the organisation that continually creates the conditions for the next category to emerge.
There is a wonderful paradox here.
The strongest technology company may eventually be the one that is least afraid of its best people leaving. Because it knows that if it has created the right environment, those people don’t really disappear from its orbit. They go and create something new. And the company can back them.
But there is a danger.
If you turn every departure into a spin-out, you can hollow out the core. At some point, you need people inside the organisation who are actually building the products. You can’t become the venture-capital arm of your former employees. That’s why Google faces a fascinating test. Can it keep Gemini and its core AI businesses moving quickly while allowing its most ambitious researchers to pursue entirely new directions?
Because category leadership still requires execution. You need infrastructure. You need products. You need distribution. You need customers. You need a commercial engine.
The two-speed technology company
Perhaps we’re therefore moving towards a two-speed model.
Inside the company – scale, products, infrastructure and execution.
Around the company – research, spin-outs, founders, moonshots and entirely new categories.
And the parent company becomes the connective tissue linking startups and scaleups.
That could be extraordinarily powerful for Big Tech because these companies possess something most venture-backed startups don’t; enormous balance sheets. Compute. Distribution. Customers. Talent networks. Decades of institutional knowledge.
So why force every ambitious researcher to choose between being an employee and being completely independent, scrambling for capital, resources and distribution?
Maybe there’s a third option of “Go build the next thing. We’ll back you.”
That changes the psychology of talent management.
Instead of asking “How do we stop our stars from leaving?”, the question becomes “How do we make our ecosystem so attractive that leaving a position doesn’t mean leaving us behind?”
The relationship can outlive the employment contract.
Broad tech
And perhaps that’s the bigger lesson from Hassabis and Dean.
One is moving further into the scientific future while remaining deeply connected to Alphabet. The other is leaving to build something new, with Google as an investor.
Two different versions of the same idea
- The relationship can survive the employment contract. That may be the real new model for achieving category leadership.
- Don’t try to own every great person. Own the ecosystem in which great people can do their best work.
Because over the next decade, the category leader might not be the company with the biggest team. It might be the company with the strongest gravitational pull around talent, capital, technology and ideas. And if that’s true, then the smartest thing a Big Tech company can do when its star researcher says, “I want to build something new”, might not be to say no.
It might be to ask “How big do you think this could become – and how can we help?”
Now that would be a very different way of running Big Tech.
A model built not just around scale, but around continually fostering innovation and creating new categories.
Perhaps we should call it Broad Tech.
#bigtech #broadtech #talentmanagement #categoryleadership