The new site-selection metric is people: AI talent density is redrawing the competitiveness map

The headline of Stanford's 2026 AI Index is not which country has the most money or the biggest models. It is that two of the smallest serious economies in the world, Switzerland and Singapore, sit at the very top of the global ranking for AI talent density, with 110.5 and 109.5 researchers and developers per 100,000 people respectively. Germany, the UK and most large economies trail well behind on a per-capita basis. The lesson is uncomfortable for jurisdictions used to competing on scale: in the innovation economy, concentration beats size.

A second finding sharpens the point. China has narrowed the model-performance gap with the United States to roughly 2.7% on leading benchmarks, while spending a small fraction of US levels on AI. Whatever one makes of the specific number, the direction is clear: competitiveness is increasingly a story about how efficiently a place converts talent and compute into capability, not how much raw capital it can mobilize. The countries climbing the index are those that have built dense, well-connected pools of skilled people and let them compound.

This is happening just as the physical build-out of the AI economy accelerates. SEMI expects advanced chipmaking capacity to grow 69% through 2028, with global output crossing a million wafers a month for the first time this year. The capital behind it is staggering: a $55 billion SpaceX 'Terafab' filing in rural Texas, a $7 billion Amkor packaging campus in Arizona, multi-billion-dollar fabs across the US, Europe and Asia. But every one of these projects is, underneath the incentives, a bet on access to engineers: process, design, packaging and increasingly AI-systems talent that is in chronically short supply.

That is the bind for economic developers. The classic competitiveness toolkit (incentives, serviced land, tax treatment) still matters, but it is no longer the binding constraint for the most strategic projects. A fab or an AI lab can be financed almost anywhere; it cannot be staffed anywhere. When a hyperscaler or a foundry shortlists locations, the live question is whether the local labor shed can supply hundreds or thousands of specialized engineers within the project's ramp window, and whether the surrounding ecosystem will keep replenishing them.

The Swiss and Singaporean playbooks are instructive precisely because neither relied on giant incentive budgets. Both built density deliberately: world-class universities tied tightly to industry, aggressive skilled-migration policy, and a small enough geography that talent, capital and firms collide constantly. Density is not an accident of size. It is a policy outcome, achievable by mid-sized regions willing to specialize rather than spread themselves across every emerging sector.

For practitioners, the strategic implication is to start measuring and marketing the thing that now decides projects. Most agencies can recite their tax rates and land costs from memory but cannot quantify their AI or engineering talent density, their graduate pipeline, or their net retention of skilled workers. In a cycle where the scarce input is people, the jurisdictions that win will be the ones that treat talent density as a core competitiveness metric, building it patiently and pitching it credibly, rather than as a soft factor in the appendix of the site-selection deck.