NVIDIA's Blackwell AI Chip Faces Overheating Issues, Causing Shipping Delays

Reportedly, NVIDIA's latest Blackwell AI chips have faced multiple challenges since their launch. These high-end chips are not only experiencing shipping delays but also overheating issues in the supporting servers, raising concerns among customers about the construction and operational timelines of new data centers.

Insiders reveal that when Blackwell GPUs are connected together within server racks, overheating occurs due to design flaws. This issue is particularly evident in server racks designed to accommodate up to 72 chips.

Further reports indicate that NVIDIA has repeatedly requested suppliers to modify rack designs to address the overheating problem. Several NVIDIA employees, informed clients, and suppliers have confirmed this information, although the specific names of the suppliers were not disclosed in the report.

In response to this challenge, a NVIDIA spokesperson stated in an interview: "NVIDIA maintains close partnerships with leading cloud service providers; they are an integral part of our engineering teams and processes. Iterations are a normal and expected part of the engineering development process. We are actively addressing the current issues and are committed to delivering the highest quality products and services to our customers."

In March of this year, NVIDIA officially launched the Blackwell chip, initially planning shipments in the second quarter. However, a series of issues have forced a delay in the shipping timeline. This delay may affect several key clients, including Meta (for certain platforms), Google, and Microsoft.

The Blackwell chip is an innovative product from NVIDIA, which integrates two silicon wafers of the same size as the company's previous products, achieving an impressive 30-fold speed increase. This performance boost significantly enhances the responsiveness of tasks such as chatbots, further strengthening NVIDIA's competitiveness in the AI chip market.

Nevertheless, the current challenges undeniably place considerable pressure on NVIDIA. Addressing the overheating issues promptly, ensuring timely chip deliveries, and meeting customer demands will be crucial challenges that NVIDIA must tackle moving forward.