Nvidia Reports Record Revenue as Blackwell Supply Crunch Intensifies
Nvidia reported a historic surge in quarterly revenue to $35.1 billion this week, shattering Wall Street expectations despite looming engineering challenges with its next-generation Blackwell AI processors. The Santa Clara-based chipmaker revealed that demand for its AI infrastructure continues to outpace its ability to manufacture chips, a bottleneck that CEO Jensen Huang confirmed will persist well into the coming fiscal year. While the company’s Data Center division grew by an astronomical 112% year-over-year, the earnings call was dominated by concerns regarding the manufacturing yields and thermal management of the new GB200 NVL72 server racks.
The announcement follows a volatile week of industry reports alleging that Nvidia’s flagship Blackwell chips were suffering from severe overheating issues when deployed in high-density server clusters. The Information reported that the new liquid-cooled racks, designed to house 72 interconnected GPUs, required multiple design iterations to prevent thermal throttling, potentially delaying deployments for major hyperscalers like Microsoft and Meta. During the earnings briefing, Huang dismissed the severity of these claims, characterizing the changes as standard engineering refinements rather than fundamental flaws. He emphasized that the company is currently shipping Blackwell systems to partners including Oracle and CoreWeave, with production expected to accelerate through the fourth quarter.
Despite the optimistic revenue figures, Nvidia’s gross margins dipped slightly to 74.6%, a reflection of the increased costs associated with ramping up the complex Blackwell architecture. The transition from the mature Hopper H100 architecture to the new platform has introduced yield challenges at the foundry level, specifically with manufacturing partner TSMC. CFO Colette Kress signaled that while Blackwell shipments will generate billions in revenue in the fourth quarter alone, supply constraints will likely limit the company’s upside potential for several months. This supply-demand imbalance forces customers to compete for limited allocations, maintaining Nvidia’s pricing power even as production costs rise.
The semiconductor giant’s forward guidance indicates a slowing of its explosive growth rate, dropping from triple-digit percentages to a projected 70% for the next quarter. This deceleration has tempered immediate investor enthusiasm, causing the stock to dip slightly in after-hours trading despite the beat on top and bottom lines. Market analysts suggest that the hardware limitations of data center power and cooling infrastructure are beginning to act as a governor on the pace of AI adoption. As Nvidia pushes the thermal envelope with its 120-kilowatt server racks, the physical realities of data center construction are becoming as critical as the silicon itself in determining the trajectory of the AI boom.
