The Tier 2 framing is the most useful contribution here. Most economic-security conversation jumps straight from full reshoring (Tier 1, CHIPS Act sized) to laissez-faire (Tier 3), and the middle is where actual money goes to die in practice. A $20 to $30 billion Latency Fund explicitly priced for option value rather than steady-state production is the kind of instrument that could survive a transition between administrations because the deliverable is months-to-substitution, not jobs-by-state. The nuclear-latency analog also disciplines the strategic question: which Tier 2 bottlenecks would you rather have priced into Beijing's calculus, and which would you rather have priced into yours?
China has been running this playbook for almost a decade through 自主可控 (autonomous and controllable) -- specifically targeting the Tier 2 middle band you describe. Loongson CPUs, Huawei's Kunpeng/Ascend, CXMT DRAM: none are competitive with TSMC/Nvidia benchmarks in peacetime, but each preserves the capacity to substitute at scale under pressure.
The mechanism isn't autarky -- it's optionality. China's new supply chain security regulations (April 2026) operationalize exactly this: Article 9 creates a formal inter-ministerial review process for foreign supply dependencies in critical sectors. The H200 non-delivery this week is that logic running in real time. Chips US-approved, companies Beijing-approved, but the 25%-transit-through-US arrangement triggered the review. 75,000 chips per approved buyer, undelivered.
Your Tier 2 framework maps cleanly onto what China calls 补链 (fill chain gaps) vs 强链 (strengthen the chain) in 十五五 policy. 补链 = latency plays. 强链 = permanent capacity. The 15th Five-Year Plan allocates separate instruments to each.
The implicit lesson for a latency fund: credibility requires a small-enough gap between latent and active capacity. China's domestic compute isn't competitive today, but it's close enough that the optionality is real and Beijing uses it. The US version needs a similar time-to-activate benchmark to function as deterrence rather than just insurance.
My essay on “The Rise of Mesoeconomics” <https://www.billjaneway.com/the-rise-of-mesoeconomics> outlines the fundamental need for empirical analysis of production networks informed by network theory to address the issues defined and reviewed in this excellent edition of China Talk. Appropriately, my essay begins with a comprehensive study of the lessons learned from full industrial mobilization in World War II and the partial mobilization triggered by the Korean War, The Economics of National Security, published by the Social Science Department of the US Military Academy in 1952.
The Chinese tech sector's behavior offers an interesting case study for this framework. The shift from Nvidia to domestic alternatives isn't driven by performance parity — it's driven by exactly the risk the article describes: latent capability to operate without an adversary's technology.
The H200 story: Chinese cloud vendors declined to purchase chips the US cleared for sale. They're paying a performance price now to avoid a dependency risk later. Nvidia's China share has fallen from 90%+ to 55%, domestic alternatives at 41%. That's the article's logic operating bottom-up rather than planned top-down.
When replenishment velocity of essential requisites falls below operational consumption and cannot be restored within strategic time horizons, the rational policy response shifts from war preparation to negotiated coexistence. Coexistence is not a concession. It is a temporal strategy for rebuilding the material conditions of power:
The Tier 2 framing is the most useful contribution here. Most economic-security conversation jumps straight from full reshoring (Tier 1, CHIPS Act sized) to laissez-faire (Tier 3), and the middle is where actual money goes to die in practice. A $20 to $30 billion Latency Fund explicitly priced for option value rather than steady-state production is the kind of instrument that could survive a transition between administrations because the deliverable is months-to-substitution, not jobs-by-state. The nuclear-latency analog also disciplines the strategic question: which Tier 2 bottlenecks would you rather have priced into Beijing's calculus, and which would you rather have priced into yours?
China has been running this playbook for almost a decade through 自主可控 (autonomous and controllable) -- specifically targeting the Tier 2 middle band you describe. Loongson CPUs, Huawei's Kunpeng/Ascend, CXMT DRAM: none are competitive with TSMC/Nvidia benchmarks in peacetime, but each preserves the capacity to substitute at scale under pressure.
The mechanism isn't autarky -- it's optionality. China's new supply chain security regulations (April 2026) operationalize exactly this: Article 9 creates a formal inter-ministerial review process for foreign supply dependencies in critical sectors. The H200 non-delivery this week is that logic running in real time. Chips US-approved, companies Beijing-approved, but the 25%-transit-through-US arrangement triggered the review. 75,000 chips per approved buyer, undelivered.
Your Tier 2 framework maps cleanly onto what China calls 补链 (fill chain gaps) vs 强链 (strengthen the chain) in 十五五 policy. 补链 = latency plays. 强链 = permanent capacity. The 15th Five-Year Plan allocates separate instruments to each.
The implicit lesson for a latency fund: credibility requires a small-enough gap between latent and active capacity. China's domestic compute isn't competitive today, but it's close enough that the optionality is real and Beijing uses it. The US version needs a similar time-to-activate benchmark to function as deterrence rather than just insurance.
My essay on “The Rise of Mesoeconomics” <https://www.billjaneway.com/the-rise-of-mesoeconomics> outlines the fundamental need for empirical analysis of production networks informed by network theory to address the issues defined and reviewed in this excellent edition of China Talk. Appropriately, my essay begins with a comprehensive study of the lessons learned from full industrial mobilization in World War II and the partial mobilization triggered by the Korean War, The Economics of National Security, published by the Social Science Department of the US Military Academy in 1952.
The Chinese tech sector's behavior offers an interesting case study for this framework. The shift from Nvidia to domestic alternatives isn't driven by performance parity — it's driven by exactly the risk the article describes: latent capability to operate without an adversary's technology.
The H200 story: Chinese cloud vendors declined to purchase chips the US cleared for sale. They're paying a performance price now to avoid a dependency risk later. Nvidia's China share has fallen from 90%+ to 55%, domestic alternatives at 41%. That's the article's logic operating bottom-up rather than planned top-down.
When replenishment velocity of essential requisites falls below operational consumption and cannot be restored within strategic time horizons, the rational policy response shifts from war preparation to negotiated coexistence. Coexistence is not a concession. It is a temporal strategy for rebuilding the material conditions of power:
https://alkoch55.substack.com/p/the-ontological-flaw-in-elite-china?r=kmlt&utm_medium=ios