H200 China quota and the tier-stratified export-control equilibrium
The formalization of NVIDIA H200 exports to China under a 50% volume cap, 25% tariff, and third-party security testing — paired with continued Blackwell-generation restriction — creates the first stable two-tier export-control regime. The structure gives both US compute-supply chains and China-AI-compute deployment teams predictable conditions to plan against.
The H200/Blackwell two-tier export structure is the first stable export-control equilibrium since the 2022 binary-restriction regime began deteriorating. The implications extend beyond the China-US bilateral into the global compute-supply landscape.
What the binary regime got wrong
The 2022-2024 binary regime tried to draw a hard line: chips above the capability threshold are blocked, chips below it are allowed. The line kept moving as chip capability scaled — every generation transition created a new round of policy uncertainty. Both US and Chinese supply chains paid a planning-uncertainty premium because the threshold kept shifting.
What the tier-stratified regime fixes
The H200/Blackwell two-tier structure stratifies by chip generation rather than by capability threshold. H200 ships to China under conditions (50% volume cap, 25% tariff, third-party security testing); Blackwell-generation chips remain fully restricted. The structural improvement is the regime becomes generation-based rather than threshold-based — when new chip generations ship, the policy update is mechanical (add the new generation to the restricted tier) rather than negotiation-heavy.
The TSMC capacity scaling that enables this
TSMC's FY26 data-center revenue at $194B with 68% YoY growth and AI-accelerator CAGR projected at 54-56% through 2029 provides the manufacturing-capacity precondition for the tier-stratified regime to work. TSMC's ability to allocate H200-class production to mixed restricted-and-unrestricted markets while reserving Blackwell-generation capacity for unrestricted markets gives the policy regime the manufacturing-side flexibility it needs.
The procurement-confidence inflection
H2 2026 compute procurement now operates under stable conditions for the first time in three years. US procurement teams can plan against full Blackwell access; China-deployment teams can plan against quota-limited H200 access; both sides can model the implications of new chip generations on their procurement pipelines mechanically. The compounding effect is that supply-chain decisions made now have higher confidence intervals, which supports larger fleet deployments and longer-term capacity commitments.
The 21-49x advantage premise
Analyst estimates that the US holds a 21-49x advantage in AI compute produced in 2026 over China assume the stratified-export regime continues operating. The regime is structurally durable through the current Administration — both sides have planning-uncertainty incentives that align with regime stability. The H2 2026 compute-competitive-landscape forecast operates from a more confident baseline than 2024-2025 forecasts could.
CSIS — DeepSeek, Huawei, Export Controls, and the Future of the U.S.-China AI Race → · CFR — China's AI Chip Deficit →