What digital sovereignty means and why it matters to investors
Digital sovereignty sounds like a policy term. In practice, it’s a procurement decision made by governments, repeated at scale, across dozens of countries simultaneously.
It means…
We won’t store our citizens’ data on servers we don’t control. We won’t train our national AI models on infrastructure owned by a foreign company. We won’t let our military, healthcare, financial and government systems depend on a cloud provider subject to another country’s laws.
Every time a government reaches that conclusion, and right now, they are reaching it in waves – it triggers spending. On data centres. On chips.
On sovereign cloud platforms. On cybersecurity. On the energy to power all of it.
The World Economic Forum forecasts investment in AI-dedicated infrastructure to grow at 10–15% annually, reaching more than $400 billion per year by 2030.
The US and China are capturing roughly 65% of aggregate global AI investment between them. The remaining 35% – the rest of the world building its own sovereign AI capacity – is where the structural investment opportunity sits. Because those countries can’t buy American chips at scale, can’t trust Chinese infrastructure, and have no choice but to build their own. Every one of those decisions flows through a relatively small set of companies.
The chip war that’s reshaping everything
Sitting at the centre of the sovereignty investment thesis is the most consequential technology battle in modern history: the US-China semiconductor war. And it just escalated significantly.
In April 2026, a bipartisan group of US lawmakers introduced the MATCH Act (the Multilateral Alignment of Technology Controls on Hardware). The bill would extend US export restrictions beyond American companies to include allied suppliers: specifically, ASML’s deep ultraviolet lithography machines and Tokyo Electron’s etching and deposition tools – equipment that Chinese chipmakers have continued to access even as the most advanced systems have been blocked since 2019.
The logic is surgical and deliberate. Chipmaking equipment isn’t static. It requires constant calibration, spare parts, software updates and field support. Cutting off the servicing doesn’t just close the door on new sales. It starts a countdown on the usefulness of equipment already installed in China. The MATCH Act, if passed, doesn’t just restrict future Chinese capability. It begins degrading existing capacity.
China was ASML’s largest single market in 2025, representing 33% of total revenue. By Q1 2026, that share had already compressed to 19%, before the MATCH Act. If the legislation passes, the decline would be far steeper. Applied Materials has already projected $600–710 million in lost China revenue for fiscal 2026.
The immediate read is negative for ASML. The structural read is more nuanced and more interesting.
Why the chip restrictions create the opportunity
Every country that can no longer buy advanced chips from China or trust that its chip supply won’t be cut off in a geopolitical dispute, now needs to build domestic capacity. That requires chipmaking equipment. From ASML. From Applied Materials. From Lam Research. From Tokyo Electron. The very companies being restricted from selling to China are the only companies capable of supplying the sovereign chip buildout everywhere else.
Europe has understood this clearly. The EU’s €700 million investment in a NanoIC semiconductor pilot line at IMEC in Leuven (backed by ASML and national governments) reflects the assessment that the chip supply chain is fragmenting and that domestic capacity is no longer optional. France is deploying 1.2 million GPUs and training 100,000 AI professionals annually by 2030 under its France 2030 plan. Germany’s public cloud spend is forecast to grow 17% in 2026 alone.
South Korea’s programme is the most ambitious of all: a $735 billion sovereign AI initiative with Samsung committing $230 billion, the government allocating $185 billion for AI research and development, and $300 billion earmarked for infrastructure. This is not a policy aspiration. It is the largest single national AI programme in history, with legal foundations (the AI Basic Act, effective January 2026) already in place.
The Gulf states are building at a different kind of scale entirely. AWS is constructing a cloud region in Saudi Arabia as part of a $5.3 billion commitment. Microsoft has committed $15.2 billion to the UAE through 2029. Google Cloud and the Saudi Public Investment Fund have announced a $10 billion AI hub partnership.
The rationale in every case is identical: sovereign nations with significant resources cannot afford to have their critical digital infrastructure owned by a foreign power, particularly after Iranian drones proved that data centres can be targeted like any other military asset.
For investors, the digital sovereignty theme has three distinct layers:
Layer 1: The Semiconductor Equipment Monopoly
ASML, Applied Materials, Lam Research and Tokyo Electron are the irreplaceable toolmakers of the chip world. Every sovereign chip programme – in Europe, Korea, Japan, India, the Gulf – runs through this equipment.
Layer 2: The Sovereign Cloud Infrastructure Build
Every government building sovereign AI capacity needs data centres. The companies building them – Equinix, Digital Realty, Vertiv, Schneider Electric – are the physical infrastructure layer of the sovereignty trade.
Layer 3: The Picks-and-Shovels Plays Nobody Is Talking About
Sovereign AI infrastructure requires power at unprecedented scale, cooling at unprecedented efficiency, and cybersecurity at unprecedented depth. The companies solving the energy constraint for sovereign data centres – power management, battery storage, grid-scale UPS systems – are arguably more structurally critical than the chip companies themselves.
The risks?
Digital sovereignty is not a frictionless investment theme. The risk that is consistently underpriced is fragmentation cost. A world of 30 sovereign AI clouds is less efficient than a world of three hyperscale ones.
Duplicating infrastructure across jurisdictions costs money – money that ultimately comes from government budgets or is passed through to enterprise customers. The productivity gains from AI may be partially offset by the cost of building the sovereignty layer required to deploy it safely.
There is also a concentration risk that runs in the opposite direction. The WEF notes bluntly that a small number of economies are pulling ahead in access to advanced chips, reliable power and high-assurance data centre capacity, and that without new infrastructure models, many economies risk missing out entirely.
Sovereignty, in other words, may end up concentrated in the hands of a dozen well-capitalised nations, with everyone else dependent on their infrastructure rather than American or Chinese infrastructure. The map changes. The dependency doesn’t.
Not a subscriber to Money Morning?
You can get free daily recommendations like these with Money Morning eletter. Just sign up here.
