Subsidize, Substitute, Secure: Three National Playbooks
When a resource becomes strategic, governments stop trusting the market to supply it and start intervening directly. In semiconductors and AI compute, that intervention has taken three broad forms, each associated with a different position on the global board. Call them subsidize, substitute, and secure. Understanding these playbooks lets you read almost any national announcement about chips or AI as a variation on one of three underlying moves.
The first playbook is subsidize, and it belongs to the nations that already sit near the frontier but want to bring more of the supply chain onto their own soil. The clearest example is the United States. Having discovered that the most advanced chip manufacturing sat overseas, concentrated dangerously in Taiwan, Washington moved to bring fabrication home. In 2022 it passed legislation committing tens of billions of dollars in subsidies and incentives to build semiconductor factories domestically. The logic is straightforward: designing the best chips is worth less if you cannot manufacture them within your own borders during a crisis. Europe, Japan, and South Korea have pursued their own versions of the same move, offering public money to attract or expand advanced fabrication at home. The goal is resilience: shortening and domesticating the supply chain so that no single foreign point of failure can sever national access to compute.
The second playbook is substitute, and it belongs to the nation on the receiving end of export controls. Cut off from the frontier by design, China has made semiconductor self-sufficiency a central national priority, directing vast state resources toward building an entire domestic stack: chip design, fabrication, and eventually the manufacturing equipment it is currently denied. Progress is real but uneven. In mature and mid-range chips the effort has advanced substantially. At the leading edge, and especially at the deepest choke point of advanced lithography, the gap remains wide and hard to close quickly. But the strategic commitment is total, because the alternative, permanent dependence on a rival for the master input of the age, is unacceptable to any great power. Substitution is slow, expensive, and driven by necessity rather than efficiency, but it is relentless.
The third playbook is secure, and it belongs to a rising tier of ambitious states that do not design or manufacture chips but are determined to possess compute nonetheless. These nations, several of them in the Gulf, bring two assets to the table: abundant capital and abundant energy. They cannot build the stack, so they buy into it, striking deals to acquire large quantities of advanced chips, funding domestic AI data centers, and investing in frontier AI companies abroad. Their bet is that in a world where compute is power, a nation with money and cheap electricity can purchase a seat at the table even without the underlying technology. The controlling nations, for their part, use access to chips as a diplomatic instrument, granting or withholding it to draw these ambitious states into their orbit rather than a rival's.
These three playbooks are not mutually exclusive, and the most capable nations run more than one at once. The United States subsidizes domestic fabs while also using chip access to secure alliances and deny rivals. But the framework clarifies the underlying motives. Subsidize is the move of the incumbent shoring up resilience. Substitute is the move of the constrained power seeking independence. Secure is the move of the ambitious newcomer buying its way in.
What unites all three is a rejection of the previous era's assumption that the global market would reliably supply whatever a nation needed. That assumption held when the resource in question was not strategic. It collapses the moment compute becomes the terrain of national power. Every playbook is, at bottom, an answer to the same question each government is now forced to ask: how do I ensure my nation has the compute it needs, on terms I control, regardless of what my rivals or even my allies decide to do?
My read is that we are watching the early innings of a long and expensive reindustrialization, driven by security logic rather than pure economics. Factories are being built in places where they would never be built on cost grounds alone, because resilience is now worth paying for. The market optimized for efficiency and gave us a fragile, concentrated supply chain. Governments are now optimizing for control, and are willing to sacrifice efficiency to get it. That trade, efficiency for sovereignty, is the defining economic decision of the compute era.