Some Scribbles.

Demystifying the Semiconductor Value Chain: Core Silicon

Surging demand for AI data centers has driven significant appreciation in the stock prices of semiconductor companies across the value chain. In this series, we'll break down Nvidia's GB200 supply chain to map which components and suppliers have captured the data center build-out, and why the market has repriced certain names along the way.

GB200 Spotlight: Core Silicon

The Nvidia GB200 Superchip is built around two Blackwell GPUs and a Grace CPU. These processing units perform trillions of calculations per second, supporting the training and inference of models that generate text, images, and other content. However, beyond chip design, Nvidia's actual advantage lies in software. PyTorch, TensorFlow, and most ML tooling hug tightly around Nvidia's CUDA stack for training and running models. Rivals can ship competitive hardware, but replicating CUDA's developer and library footprint is a different problem.

As AI model companies depend on Nvidia's chips and software, TSMC manufactures the chips Nvidia designs. In particular, Nvidia relies on TSMC's proprietary way of packaging a GPU and its memory (HBM) side by side with fast, short connections between them. This packaging and fabrication process is called CoWoS, and only TSMC's Taiwan fabs offer it today. Without CoWoS, models with up to trillions of parameters would struggle to achieve the performance levels required for commercial deployment.

CoWoS Packaging Comparison

Nvidia's data center revenue grew every quarter over the past two fiscal years, climbing from $22.6 billion to $62.3 billion:

Nvidia Datacenter Revenue

Demand reached a clear inflection point in Q3 FY2026 as cloud providers raced to secure GPU capacity while supply tightened across the semiconductor supply chain. The GPU itself is only part of the story. The GB200 platform relies on a complex network of suppliers providing memory, networking, packaging, power, and cooling. Each captures a slice of the same hyperscale buildout that drove the revenue curve as shown above. In the coming installments of Demystifying the Semiconductor Value Chain, we'll work through those surrounding suppliers and bottlenecks, and the role each plays in the value chain.