Nvidia has announced that it will sell more of its next-generation Blackwell chips than initially projected, as the company continues to see strong growth in its AI and server businesses. The company’s recent quarterly sales nearly doubled, sparking investor questions on how long Nvidia can sustain such rapid expansion with its $140 billion annual revenue run rate. Much of Nvidia’s future growth is tied to the success of Blackwell, a new family of server products designed around its cutting-edge AI chip.
CEO Jensen Huang and CFO Colette Kress addressed these concerns in a call with analysts on Wednesday, where they provided new insights into Blackwell’s production and rollout. According to the executives, the Blackwell launch is progressing as planned, with Nvidia’s production capabilities being the primary limiting factor in sales rather than demand. Huang stated that Blackwell production is “in full steam” and confirmed that the company will ship more units this quarter than previously expected. Despite missing some overly optimistic expectations from investors, Nvidia’s positive outlook on Blackwell sales helped its stock drop by only 1%.
A key update from the call was the confirmation that Nvidia had already shipped 13,000 Blackwell chips to some of its major customers, including Microsoft, Oracle, and OpenAI. These companies have showcased the chips in their server racks on social media, highlighting the growing importance of Nvidia’s products in powering next-generation AI systems. However, the initial shipments were primarily early versions of the Blackwell chip, designed to allow customers to begin testing and prepare their systems for the volume shipments expected in the coming months.
Nvidia’s management expressed confidence that sales of Blackwell would continue to ramp up over the next few quarters. Huang noted that the company expects to ship more units next quarter than in the current one and to continue increasing shipments thereafter. In July, Nvidia had projected “several billion dollars” in Blackwell revenue for its current quarter, and it has since raised that expectation.
While demand for Blackwell is high, Nvidia has faced supply chain challenges. Huang acknowledged that the company’s suppliers have struggled to meet the immense demand for the new chips. He emphasized that the generative AI revolution is driving this surge in demand, with companies like OpenAI urgently requiring the fastest GPUs available to build new AI models.
Nvidia also discussed Blackwell’s expected gross margins, forecasting that they will be lower than the 73.5% seen in the third quarter, but will improve as production scales. The company’s broader strategy is to position Blackwell as the go-to chip for the next generation of AI models, eventually surpassing its previous AI chip, Hopper, in sales. Huang pointed out that while Hopper was used in the final stages of the last generation of foundation models, Blackwell would be the starting point for the next wave of AI development, with initial sales expected to rival the scale of Hopper’s sales.
Overall, Nvidia’s Blackwell rollout appears to be on track, with the company continuing to solidify its dominance in the AI chip market, despite facing production and supply chain challenges.
