Skip to content
GLOBALANCE INTELLIGENCEThe economy ahead.
This Week Index 28/100 Sign in
Future 2040 › The 14 pillars
2040 outlookThe 14 PillarsWhat Can Go RightThe Big Questions
← Back to the pillars
Systemic competition · Pillar 11

Whoever controls the AI infrastructure controls the next economic cycle.

In the economy ahead, the classic location advantage — cheap labour — loses its relevance. Tech sovereignty takes its place: who owns the chips, the models, the data, the energy? This shift forces companies, states and investors to fundamentally reassess location risks. For Switzerland as an open economy situated between three geopolitical blocs, this is an existential question.

Pillar 11KI
01 · Status quo & fault line

Where do things stand today, and where do they break?

Status quo

The world is undergoing an accelerating deglobalisation. The US CHIPS Act (2022) mobilises $52bn for domestic semiconductor production. The EU AI Act and Digital Markets Act regulate access to technology. China's Made in China 2025 strategy and its response to Western sanctions make clear that AI is not merely an economic phenomenon — it is a geopolitical instrument of power. Trade volumes between G7 countries and China have been structurally declining since 2017. Nearshoring: reshoring announcements in Europe rose 34% between 2022 and 2024. The CSIS technology-decoupling index shows technology transfer between the United States and China at a historic low in 2024.

Fault line

AI renders traditional locational advantages obsolete: Labour Costs become irrelevant when automation dominates. Scaling based on cheap labour in low-wage countries falls away. What counts instead is access to chips, energy infrastructure, the regulatory environment, AI talent and data sovereignty. This creates a new geopolitics of location. Countries without their own AI infrastructure will become structurally dependent, much as oil importers were in the 1970s. For Switzerland, this means dependence on NVIDIA chips (US-controlled), energy pressure from data-centre expansion and the question of whether «neutrality» between tech blocs remains a credible strategy.

02 · Policy options

Paths out of the crisis

The stars show how realistic a path is: 1 star = barely feasible, 5 stars = highly feasible.

A★☆☆☆☆ 1/5

Full technological sovereignty

Own chips, own data centres, own foundation models: the entire chain kept within the country.

+ Independent of third-party export controls.

The scale is beyond the reach of an economy this size: a single cloud provider invests more in data centres each year than most countries spend on research in total.

Trade-off: Ties up capital that would generate greater impact elsewhere.

Feasibility: Popular as an aspiration, yet delivered nowhere as a programme. Even the major chip programmes pursue market-share targets, not autarky.

B★★★☆☆ 3/5

bloc alignment

A clear alignment with one technology bloc: access in exchange for allegiance.

+ Access to leading-edge chips and frontier models remains secured.

The bloc's sanctions become its own. Its negotiating position falls to zero — permanently.

Trade-off: Security versus freedom of action.

Feasibility: The most convenient path and the hardest to reverse. Once you are locked in, you no longer have any leverage.

C★★★★☆ 4/5

Diversification

Multiple sources for chips, models and computing power — with no exclusive lock-in to any single vendor.

+ Retains freedom of action and negotiating leverage.

Costs more and requires foreign-policy capacity that must be sustained on an ongoing basis.

Trade-off: Cost versus optionality.

Feasibility: The classic response of small open economies. It works as long as more than one provider exists.

D★★★★☆ 4/5

Cooperative Capacity

Computing and model infrastructure built through European collaboration rather than individual national effort.

+ Scaling without overstretching national capacity. And standards are co-written rather than adopted.

The decision-making pace of a consortium is pitted against product cycles measured in months.

Trade-off: Reach versus speed.

Feasibility: This works where infrastructure is long-lived: more so for data centres and networks than for models.

E★★★★☆ 4/5

Specialisation over sovereignty

Not by occupying the base infrastructure, but by holding the layer above it — one that cannot be relocated.

+ Realistic for a small open economy: precision industry, life sciences, financial infrastructure, legal certainty.

Offers no protection against the failure of the base on which the specialisation rests.

Trade-off: Indispensability at the top against dependency at the bottom.

Feasibility: The path small countries have historically taken, and the only one that works without a size advantage.

03 · Time horizon

Trigger-based phases

Phase 1
Control of supply chains

Trigger: Export restrictions on chips and critical raw materials are becoming the norm

Access to technology is becoming the currency of foreign policy. Those who supply it demand commitments in return. Location decisions are no longer driven by labour costs but by the availability of chips and energy.

Phase 2
Two technology spheres

Trigger: Separate standards for models, data and certification

Products must be built to satisfy two regulatory regimes. For export nations, costs rise; for intermediaries, value rises. Neutrality becomes expensive and, for that very reason, negotiable.

Phase 3
A new location logic

Trigger: Energy and legal certainty overtake labour costs as a location factor.

Value creation concentrates where power is predictable, the legal framework is robust, and talent is available. Countries lacking all three become structurally dependent — much as energy-importing nations were in the 1970s.

04 · Pilot Programmes

Real-world pilot programmes

Where the ideas behind this pillar are already being tested. "Learn more" opens the explanation without leaving the page.

EU Chips Act ↗Status: In force since 2023A European programme to build semiconductor capacity. What is notable is the objective: a share of global production — not independence.
US CHIPS and Science Act ↗Status: in force since 2022 (signed 9 August 2022)Subsidies for domestic chip production, tied to conditions: recipients may expand production in China and other "countries of concern" only within tight limits for ten years. Industrial policy and foreign policy combined in a single piece of legislation. Since 2025, the government has also converted grants into equity stakes, including a 9.9% stake in Intel (August 2025).
EuroHPC Joint Undertaking ↗Status: OngoingPooled computing capacity: member states share supercomputers rather than each building their own. The template for cooperative capacity.
Gaia-X ↗Status: Since 2019A European attempt at shared data infrastructure. And an instructive case in how much speed a union sacrifices.
05 · Early Indicators

What we are watching

Investment in chip fabrication plants within the western bloc.
Today
just over two thirds
Threshold
below 55 per cent

Once the share falls below this threshold, the chip supply chain is no longer western-dominated. Access then becomes a matter for negotiation rather than something taken for granted.

Track in the index →
Critical raw materials subject to Chinese export controls.
Today
more than 20 materials
Threshold
more than 25

Every expansion is a lever. Beyond that threshold, vulnerability can no longer be bridged by stockpiling.

Track in the index →
New bilateral AI agreements between states
Today
growing
Threshold
more than 30 in 18 months

A patchwork of bilateral agreements is replacing the world order. For small countries, this means negotiating rather than joining.

Track in the index →
06 · Globalance Position

Our assessment

At Globalance, we regard full-scale technological sovereignty as a goal an economy of this size can neither achieve nor needs. The competitive advantage of the next twenty years will come not from autarky, but from indispensability.

A sustainable approach combines three paths: Diversification of supply sources, collaborative capacity within the European framework for everything long-lived (data centres, networks, certification) and Specialisation in those layers where precision, life sciences, financial infrastructure, and legal certainty are decisive. None of the three paths is sufficient on its own.

Against bloc alignment we advise against, even though it is the most convenient option in the short term. It trades access for negotiating position, and that trade is virtually irreversible: a country that has once become part of a technology ecosystem no longer negotiates the terms — it receives them.

From a Swiss perspective there are four levers a small country genuinely controls: the availability and price of energy for data centres, the predictability of regulation, access to talent, and a legally sound data location. Chips are not among them. Whoever narrows the debate to chips is negotiating over the one thing they do not control.

What this means for investors: First, data centres are built where power, connectivity and cooling are available. Energy suppliers, grid operators and providers of waste-heat recovery benefit from that expansion. Second, European computing and certification infrastructure receives public support; those who help shape the standards early secure contracts. Third, a data location that holds firm even against foreign access demands is a selling point for Swiss providers of cloud and data services. Fourth, dependence on a single provider or legal jurisdiction is a concentration risk, for companies and portfolios alike.

Why now: Export controls, chip allocation, and model access will be decided in the years ahead, not in twenty. Those who fail to build negotiating leverage in this phase will negotiate without it later.

The case against

The strongest objection is that specialisation presupposes dependency rather than replacing it. An indispensable layer is worthless if the base beneath it can be switched off. Export controls act precisely there. Relying on diversification also assumes that more than one provider will exist over time; for leading-edge chips and leading-edge models, that number has not grown in recent years. And cooperative capacity has a speed problem that cannot be organised away: a union decides in years, while technology moves in quarters.

07 · Open Questions

What we do not yet know

  1. What does neutrality mean when the foundational technology infrastructure sits in two separate spheres?
  2. Which layer of the value chain can a small country occupy in such a way that it cannot be relocated?
  3. How much energy is a country willing to make available for data centres, and at what price for all other consumers?
  4. Who negotiates chip access for countries without domestic production: the state or companies?

What comes next