Deepening capability means hiring ahead of the charter

There's a decision that arrives for every centre once it passes roughly 800 people: keep scaling the work you already run, or start deepening what the centre is actually capable of owning. Scaling is the path of least resistance. It's measurable, fundable, and easy to defend in a quarterly review, another 50 seats against another tranche of the same work. Deepening is harder to put on a slide, so most centres put it off.

Deferral has a cost that shows up late. A centre that only ever scales stays a delivery arm, however large it gets, and delivery arms are the first line a parent rationalises when the cycle turns. A centre that deepens becomes hard to move, because the capability now lives there and nowhere else. Hiring precisely to your present mandate looks disciplined on paper, no idle seniority, every role justified by work in hand. It's also, quietly, a choice to stay where you are.

The two paths look similar for a while and then diverge sharply. The scaled centre and the deepened centre can be the same headcount, the same cost line, the same org chart at a glance. The difference is what happens when the parent asks the centre to own the judgment rather than run the process. In the scaled centre, there's no one already inside who was hired for that, and the charter can't move until a 2- or 3-quarter hunt concludes with the parent watching.

Deepening, moving from executing decisions to making them, is a talent state before it's an org state. The centres that climb the value curve put the senior owner in the building ahead of the mandate, so that when the parent is ready to hand over judgment, the person who can hold it is already trusted, already networked into the group, already credible on a call the centre used to be absent from. Capability leads charter.

The playbook the pulling-ahead centres run is unglamorous. They now hire for ownership: leaders who can carry an outcome end to end, and specialists deep enough that the parent trusts them with the hard version of the problem rather than the routine one. Then they re-shape the surrounding bench so those people can actually operate, instead of dropping a principal into a team built to take instructions.

It looks expensive next to a scaling plan of the same size, because you're paying for judgment instead of hours. But it's the spend that changes what work the parent is willing to send. Within a year or two, the deepened centre is doing things the scaled one structurally cannot, and that gap compounds, because each new charter makes the next one easier to win.

Once you accept that deepening is a hiring problem, the next question is how you source each piece of it, and here another old habit is retiring. For a long time the sourcing model was a centre-level stance: this GCC builds in-house, that one leans on partners. What we're watching now is that stance dissolve into a per-role judgment. Within the same team, one capability is built permanently, the next is bought as a ready-made pod, the one beside it is borrowed through a contract specialist for as long as it's needed. The unit of decision has moved from the org to the seat.

The reason is that the three options answer different questions. Building owns the muscle but costs time and commitment. Buying skips the learning curve, though you inherit someone else's way of working. Borrowing gets you moving today and leaves nothing behind when it ends. No single policy is right across a whole charter, because different roles genuinely want different answers, and pretending otherwise leaves either speed or ownership on the table. The best centres make this call 30 times a year, not once.

The per-role call isn't a fixed calculation, because the line it turns on keeps moving, and right now it's moving in one direction for the senior end of scarce skills. For years the default in AI and data was to buy scarce capability off the market: hire the specialist, pay the premium, plug the gap. Now the leading centres are flipping it, building that capability in-house at the top, growing their own principals and applied leaders rather than renting them at ever-rising prices.

The problem is that most workforce plans encoded last year's answer. They still assume you buy the senior AI talent and build the junior layer, exactly backwards from where the leading centres are heading. When the line moves this fast, a plan that was right 18 months ago is quietly setting you up to lose, because it points your budget at buying the very roles you should now be growing.

Building capability at the top flips which roles you hire and which you grow. It means hiring for potential and adjacency, people who can grow into the scarce role, while buying selectively for the specific gaps you genuinely can't grow fast enough. The mix of “hire it” versus “grow it” is different for almost every role than it was a year ago, which is why treating it as a settled policy fails. Centres that keep re-asking the question win twice: they pay less for scarce talent, and they build the loyalty that renting never buys.

Hiring ahead of the mandate is a bet you take on purpose, and it isn't free. A senior hire made ahead of the work is, for a while, more capacity than the current mandate needs, which is exactly why most centres don't make it. It asks a leadership team to carry a little visible slack in exchange for the right to compete for the next charter at all. The alternative only moves the cost to later, as a scramble under time pressure while the parent decides whether this centre is the one to trust.

The discipline is choosing which capabilities to get ahead of. Pick the one or two areas where the charter is most likely to deepen, usually where the parent already leans on the centre for the hard version of a problem, and seat the senior judgment there early. Let the rest stay matched to today's work. A centre that tries to run ahead everywhere just carries slack it can't justify; a centre that runs ahead in the right place buys the next charter before the hunt would have even started.

For each of those bets, run the sourcing test honestly. If the capability is core and you mean to own it in three years, build it: hire permanent, absorb the ramp, keep the muscle inside. If it's real but not core, buy a mature function whole or borrow a scarce, temporary one, and know which you're doing before the req opens. Get this wrong and you discover, a year later, that you built what you should have borrowed or rented what you needed to own. The senior owner who's briefly under-utilised is the price of the next charter. Most centres won't pay it, and that's precisely why the ones who do pull ahead.