Nvidia says it's seeing “strong” demand from China for its H200 chips but is awaiting approvals from both Washington and Beijing before sales could begin
Context & Ripple Effects
Chinese interest was already pressing against supply: Nvidia was evaluating added H200 capacity after orders exceeded current output, while reports said it had approached TSMC to ramp production for prospective Chinese demand. This report makes regulatory clearance—not just manufacturing—the immediate gating factor.
The approvals question later moved in stages, from a formal US green light for H200 sales to China's first approved import batch. That sequence shows that access depends on separate decisions in both capitals, even when customer demand and supply planning are in place.
First-order effects
- Nvidia cannot convert reported Chinese demand into H200 revenue until Washington and Beijing each authorize the relevant sales and imports.
- Chinese customers seeking H200 capacity remain unable to finalize deployments or rely on delivery timing while the two approvals are pending.
Second-order effects
- Nvidia's production planning becomes harder to align with demand: earlier plans to ask TSMC to increase H200 output may need to account for an uncertain China-eligible allocation.
- Chinese buyers have an incentive to preserve alternative compute plans until import clearance is known, while Nvidia must manage scarce H200 supply across markets.
Third-order effects
- If dual approval remains the norm, high-end AI-chip market access will be shaped as much by export and import policy as by chip performance or available fabrication capacity.
- The pattern reinforces a bifurcated AI-infrastructure market in which suppliers and customers design around country-specific access rules and uncertain cross-border supply.
The trend: AI compute is becoming a policy-constrained supply chain, with national approvals increasingly determining where leading chips can be deployed.