The New Geography of Work: Location Strategy as Portfolio Design

I read Tim Marshall’s The Power of Geography in 2022, and it changed how I think about locations and long-term economic growth. The lesson that stayed with me is that geopolitics rarely announces itself. It does not arrive as a single event on a single day. It works slowly, through visa rules, trade routes, energy contracts, university enrolments and industrial policy, and it consistently shapes where capability evolves. By the time a location looks obviously attractive, the conditions that made it attractive were set in motion long before anyone put it on a shortlist.

More recently, Jamie Dimon described geopolitics as the biggest unknown facing enterprises today — a moving tectonic plate beneath the global economy. That framing matches what we are seeing in talent markets. The models enterprises have traditionally used to choose locations were not built to answer the questions now being put to them: what happens to this delivery centre if immigration policy tightens, if tariffs land on a category we ship, if the government of the country we depend on decides that AI capability is a sovereignty issue rather than a procurement one.

That is why we built The New Geography of Work, a study of how geopolitical forces are likely to reshape global talent hubs over the next five years. We referenced many published datasets in addition to Draup’s own, and the analysis highlights several clear patterns from a geopolitical lens. None of them is a forecast of a single event. They are directional shifts that a workforce planning team can begin to reason about now.

The U.S. turns inward, and Middle America gets a second look

Start with immigration. The first-order effect of tightened U.S. immigration policy is not a shortage; it is a redirection. When importing talent becomes slower and less predictable, the alternative is to build it at home. That may push companies toward home-grown talent reinvestment, accelerating hiring and reskilling across Middle America, the Midwest and the U.S. South.

If it plays out that way, it is a meaningful change for anyone whose U.S. footprint is concentrated in a small number of coastal metros. Building capability in these markets is a pipeline exercise rather than a hiring one, and a pipeline has to be started before the demand arrives. The enterprises that benefit will be the ones that started their reskilling pipelines before the policy pressure made it urgent.

Latin America rebalances rather than expands

Mexico remains a critical nearshore extension of the U.S. operating perimeter. Brazil is the more interesting case: it is emerging as a multi-axis resilience hub, with an increasing emphasis on sovereignty.

Venezuela deserves a note of caution, because it is the kind of story that invites overreaction. Political and economic shifts there may trigger renewed investment. But with nearly half of Venezuelan talent already distributed across Latin America, the impact is more likely to be regional talent rebalancing than net-new supply entering the market. If you are modelling Latin America, model the movement of people who are already working, not the arrival of a new pool.

India holds, because services-led work is insulated

India continues to dominate as the Global Capability Center hub, and the tariff conversation has not changed that. Tariffs act on goods, and services-led work has remained largely unaffected by the tariff-related disruption. If anything, the disruption has reinforced India’s strategic position, because it demonstrated the difference between a manufacturing dependency and a capability dependency.

The related movement is quieter. Quiet supply-chain rewiring is underway across enterprises. Vietnam and India’s manufacturing hubs are positioned to benefit from that rewiring even with U.S. tariffs still in play, which suggests the rewiring is being driven by risk concentration rather than by landed cost alone.

Sovereignty becomes a talent variable

Geopolitical pressures are driving AI self-sufficiency. Europe is putting increasing focus on Europe-centric foundational models and sovereign AI ecosystems. China is doubling down on homegrown models and aligning their development closely with domestic demand; the Nvidia chip delivery tensions are partially resolved, but the structural friction has not gone away.

For a workforce planner, sovereignty is not an abstraction. It can determine which models your engineers are permitted to build on in a given jurisdiction, which in turn shapes the skills you hire for, where the research roles sit, and which vendor ecosystems your people need fluency in. If model development localises, so does the demand it creates — and capability built around one ecosystem may not transfer neatly to another.

Europe buys depth; Eastern Europe picks up volume

Germany and the UK remain high-quality talent hubs, supported by historically proactive migration strategies. That said, early signs of political and social backlash are emerging in both, and it would be complacent to assume that the current openness is permanent.

Underneath that, R&D investment across Europe is rising. It may not yield immediate returns, but it positions the region well for long-term innovation and depth of capability — which is a different kind of bet from the ones a conventional location model is built to evaluate, and it should be judged on a different clock. Meanwhile, Bulgaria, Latvia, Romania and Poland are gaining traction as shared services and delivery hubs, and that is driving localised demand growth.

Canada may benefit from the same set of forces. Stricter U.S. student-visa scrutiny and shifting sentiment could push international student demand northward. That would matter more than it first appears, because student intake is where a graduate talent pool starts. A shift in where people choose to study reaches the hiring market only after they finish, which makes enrolment one of the earliest signals a workforce planner can read and one of the easiest to ignore.

Trust and neutrality as an export

The UAE continues to attract engineering and medical talent from India, supported by strong bilateral ties, targeted visa reforms and sustained demand across healthcare and infrastructure. Singapore continues to compound its advantage as a neutral, high-trust hub.

Separately, our analysis points to trusted governance hubs with growing AI talent potential, supported by strong university ecosystems and curriculum depth — a category worth watching in its own right, because governance credibility and a deep university pipeline are slow to build and hard for a rival location to copy quickly.

These are examples of the same underlying trade. When the operating environment is unpredictable, predictability itself becomes a competitive good, and locations that can credibly offer it may command a premium over locations that are merely cheaper.

Design a portfolio, not an optimum

One further pattern cuts across all of the above. As AI optimises human labour requirements, the next five years are likely to favour smaller, specialised talent hubs over continued concentration in mega-hubs. The economic logic that justified very large centres weakens when the same output can be delivered by a smaller group with sharper skills, and scale stops being the thing a site is chosen for.

The conclusion I would draw is straightforward. Location strategy is no longer an optimisation exercise; it is a portfolio design problem. The enterprises that perform best through 2026–2031 will be those that deliberately balance three kinds of sites: scale hubs that carry volume, resilience hubs that exist so that no single geopolitical event can stop delivery, and specialised micro-hubs that hold scarce capability close to a market or a research ecosystem.

Practically, that means doing three things this planning cycle. Classify every existing location into one of those three roles and be honest when a site is doing none of them well. Stress-test your two largest concentrations against a specific policy shock rather than a generic one. And start the reskilling pipelines in the markets you expect to grow into, because the lead time on capability is longer than the lead time on policy.