The case for reading every headline about chips, rare earths, cyber, and payments as one structural lens — and what it changes for builders and investors.
The US and China appear to be fighting separate battles: chips, rare earths, shipping, AI. They are not. They are manifestations of the same structural collision. If you build products or allocate capital, reading them as one collision instead of a dozen disconnected disputes changes which bets make sense.
Two systems, different rules
The US system organizes power through distributed private capital and alliances. Capital allocation sits mostly with private markets, technology lives inside companies that operate beyond the direct control of the state, security runs through a network of allies, and the dollar’s reserve status lets the whole arrangement finance itself. Decisions are spread across thousands of actors optimizing for their own returns, and the system’s strength is that there is no single failure point.
China’s system organizes power through concentrated state direction. Capital, industrial policy, and infrastructure investment are coordinated from the center, and manufacturing depth is treated as a strategic asset rather than a cost to be offshored. The state can point the entire apparatus at a chosen sector on a timeline no market would tolerate, which makes the system slower to correct and faster to mobilize.
The detail that makes this collision hardest to read is that both sides are converging in practice even as they diverge in ideology. The US is doing what it criticized China for: state-directed industrial policy, equity stakes in private companies, blacklisting commercial entities for strategic reasons. China is doing what critics said centralized systems could not sustain: producing frontier AI models, scaling memory manufacturing, building alternative financial infrastructure. Both machines are learning from each other, which is what makes this structurally durable rather than easily resolved. That convergence is why the fight is occurring on every surface area where these two systems meet.
One collision, many surfaces
When two states are at war, they contest one battlefield. When two colliding systems collide, they contest every dependency one has on infrastructure the other controls, which is everywhere.
Take semiconductors. The US built an export-control strategy on the assumption that denying advanced chips would cap China’s AI progress. In December 2024, DeepSeek trained DeepSeek-V3 on restricted H800 hardware using roughly 2.8 million GPU-hours, a result that complicates the question of whether hardware access is the binding constraint. Architecture and engineering can partially substitute for constrained chips. What stays open is whether that substitution holds at frontier-scale inference and sustained research iteration. The US answer was to pull supply onshore, with TSMC’s Arizona Fab 21 reaching volume 4nm production against $165 billion committed, while Washington warned in May 2025 that using Huawei’s Ascend chips likely violates export controls. Both systems are now doing the same thing from opposite directions: building the capability at home and denying it to the other.
Rare earths show the dependency running the other way. China refines about 91% of the world’s rare earth elements and produces roughly 94% of permanent magnets, and in April 2025 it turned that position into policy, requiring export licenses for seven heavy elements. The effect was immediate. US yttrium imports from China fell from 333 tons in the eight months before the restrictions to 17 tons in the eight months after. By October the list had expanded to twelve elements, with controls extending into supply chains that never physically touch China. The US response has been to build, with the Pentagon taking a 15% equity stake in MP Materials through a $400 million investment, though domestic production still covers only about a third of consumption.
Cyber is where the collision stops being commercial. Salt Typhoon infiltrated major US telecom networks and reached the call records and some communications of senior officials, which is classic espionage. Volt Typhoon is a different posture, pre-positioned inside US power grids, water systems, and transportation networks. US officials assess the activity as preparation that could enable disruption in a contingency. Operational intent remains contested, but the access itself is documented. I spent years working on biosurveillance and financial-sector cybersecurity programs where this pattern was visible before it had names, and the lesson that stayed with me was that the infrastructure creating commercial efficiency and the infrastructure creating strategic exposure are the same. The threat is indifferent to how the owner categorizes it.
Payments are the quietest surface and yet the most consequential. The dollar’s role in clearing global trade is the foundation of US financial power, and China’s mBridge platform is the first working piece of architecture built to route around it: a cross-border settlement system constructed with Hong Kong, Thailand, the UAE, and Saudi Arabia that reached minimum viable product in 2024 and clears transactions without USD or SWIFT. The BIS stepped back from the project in October 2024, which can be read as Western risk management working or as the architecture developing outside multilateral oversight. Either way, rails to settle international trade outside the US financial system now exist, which has been the hardest part.
Each of these surfaces has the same tell: the leverage sits in infrastructure that private companies own, which is why both governments have stopped treating it as private.
Public and private are merging
The layer most people miss is that this collision runs through infrastructure that private companies own, and governments on both sides have drawn the same conclusion: strategic competition now runs through balance sheets and product roadmaps.
The outcome depends heavily on where the middle aligns. The EU, India, Japan, Korea, and the Gulf are not passive bystanders; their sourcing, financing, and regulatory choices will determine more about which dependencies prove durable than anything either the US or China does unilaterally.
Washington has started reaching into decisions it used to leave to the market: directing where fabs get built, taking equity stakes in mining companies, deciding which drones can fly, and listing shipping lines as military entities. Beijing never needed to make that transition. Civil-military fusion is stated policy, the national security law obligates firms to cooperate, and rare earth export controls are economic policy deployed as a strategic weapon.
The most concrete version of this I have watched play out is in manufacturing and distribution: businesses making supplier and sourcing decisions that would have been purely commercial five years ago but are now effectively foreign policy calls. Tariffs changed the frame. The question shifted from who is cheapest to which of these dependencies we want to own. Most companies had never asked that question before.
Companies are becoming geopolitical actors whether or not they want the role, and a critical infrastructure firm sitting on a strategic control point does not get to opt out by declaring itself neutral.
The control points
For builders and investors, the useful frame is control points: the places where one system’s operation depends on infrastructure the other controls. What makes this different from ordinary great-power competition is that the hegemonic contest routes through private infrastructure and commercial balance sheets, not only military deployments or diplomatic channels. Compute, payment rails, logistics software, industrial process control, cyber defense, satellite communications, critical mineral processing, and defense-adjacent manufacturing are all control points, and their value is changing for reasons that have little to do with the usual growth story.
A control point is not a permanent moat. Using it invites substitution; owning it invites retaliation and regulation. The companies that benefit are not simply the ones that sit on the node; they are the ones with qualified alternatives, deep integration, and supply chains that can survive a policy shift. The non-obvious layer is one level over from the obvious one: not critical mineral processing as a category, but the separation chemistry, the assay labs, the trained workforce, the recycling feedstock that makes processing possible.
For builders, the question is whether the product sits in the flow of a critical workflow or alongside it. Products embedded in the flow accumulate a kind of strategic value that adjacent products do not, and as governments increasingly use procurement and regulatory authority to shape which products survive, that distinction carries more weight than it did five years ago.
For investors, I think the relevant question is whether a portfolio company benefits from structural demand created by security, resilience, and the drive toward strategic autonomy, or whether it depends on a low-friction global order that may be weakening. The companies building infrastructure for a world where supply chains are regionalized, data sovereignty is enforced, and allies need interoperable systems that exclude adversary components have a demand tailwind that is largely independent of the economic cycle. The companies optimized for a world where the cheapest global supply chain always wins are running against that tailwind. That is the thesis, and the honest next step is naming where it breaks.
Where I could be wrong
The core assumption here is that both systems will sustain the organizational will and economic capacity to compete across all these domains simultaneously. That could be wrong in either direction. The US has significant domestic political constraints on sustained industrial policy: the CHIPS Act passed once; getting follow-on funding through a divided Congress is a different question. China has real demographic headwinds and debt overhangs that could force priority choices between domestic stability and external competition. A negotiated partial accommodation that stabilizes some surfaces while competition continues on others is possible and would substantially change the investment thesis.
The conflict frame can also be over-applied, which comes with a failure mode. Not every supply chain decision is a geopolitical decision. Not every Chinese technology company is an instrument of state policy. The pattern I am describing is real, but applying it too broadly produces a mirror image of the naive globalization thesis: another framework that mistakes one pattern for the complete explanation of how the world works.
This framework also addresses competition below the kinetic threshold. A Taiwan contingency would not simply stress-test the investment thesis; it would temporarily suspend the commercial logic that makes the thesis investable at all. That is a different category of risk, worth naming separately.
A specific test: if US allies broadly reject export-control alignment, if bilateral trade volumes recover to 2022 levels net of transshipment effects, and if dollar invoicing share holds flat through 2027, the collision frame is weakening rather than intensifying. Those are the three indicators to watch.
The operating lens
The companies and investors who will make different decisions over the next decade are not the ones who follow geopolitics most closely. They are the ones who have updated the assumptions embedded in the prior global order: that the cheapest supplier always wins, that technology is politically neutral, that infrastructure is a commodity anyone can access. Those assumptions held long enough that they became invisible defaults.
The practical screen I now run on any investment or product question: does this depend on a stable, low-friction global order continuing, or does it benefit from a world where that order is being renegotiated? In practice, that has meant passing on deals in the last eighteen months that I would have taken before, because the business had no clear answer to which side of this renegotiation it sits on. It has also meant moving toward policy-driven and infrastructure-adjacent companies with structural tailwinds rather than products in search of a market.
The world that made the old defaults reliable is being rebuilt on different terms. Understanding which side of that renegotiation you are on is the prerequisite, and most of the world is still underwriting the old map.