Nvidia ships first H200 chips to China, but forecasts no data-centre computing revenue

Nvidia began shipping its H200 data-centre processors to China, marking a shift from the previous no-sales policy. However, these shipments only accounted for less than 1% of the company's US$89 billion data-centre revenue in its second quarter, ending July 26. Despite the limited contribution, Nvidia maintained an optimistic outlook on global demand, forecasting a 70% revenue growth for its 2028 financial year.

CEO Jensen Huang stated that actual demand was significantly higher than 70%, with the company's available supply allowing for a more substantial growth projection. The company's third-quarter revenue forecast remained unchanged, assuming no data-centre computing revenue from China. Nvidia is actively assisting AI companies in financing the infrastructure needed to purchase and deploy its chips, a move that has raised concerns about "circular financing."

The company had invested nearly US$50 billion in frontier AI laboratories and partnered with various financial institutions to mobilize over US$500 billion of third-party capital for AI infrastructure. Nvidia's shares rose by 4.71% in after-hours trading following the earnings results. Additionally, the company announced a major expansion of its partnership with Amazon Web Services (AWS), deploying an additional 2 million Nvidia GPUs over the next two quarters.

Nvidia's upcoming Vera Rubin platform, expected to account for about 20% of data-centre revenue in the third quarter, is anticipated to be the fastest product ramp-up in the company's history. However, supply constraints are projected to persist until at least the end of the 2028 financial year, and sharply higher memory prices could pressure profitability, with gross margins expected to decline from 74% in the third quarter to between 71 and 72% in the fourth quarter, before recovering to 72 to 73% in the 2028 financial year.