The Nvidia Trade: How Long Can the AI Chip Supercycle Last?

Nvidia's ascent from gaming GPU maker to the infrastructure backbone of the AI revolution is one of the most remarkable corporate stories in recent memory. With data center revenue growing from $3.8 billion in Q1 2023 to over $22 billion in Q1 2024, the company has become the defining investment of the AI era.

The Demand Picture

The immediate driver of Nvidia's growth is hyperscaler capex. Microsoft, Google, Amazon, and Meta have collectively committed to spending hundreds of billions on AI infrastructure over the next several years. Nvidia's H100 and H200 GPUs are the preferred compute substrate for training large language models, and the company's CUDA software ecosystem creates significant switching costs.

Supply Constraints and Competition

For much of 2023 and early 2024, demand for Nvidia GPUs far exceeded supply, with lead times stretching to 12+ months. This supply constraint has begun to ease as TSMC has ramped production capacity, which may moderate the pricing power Nvidia has enjoyed.

On the competitive front, AMD's MI300X has gained traction as a credible alternative, particularly for inference workloads. Custom silicon from Google (TPUs), Amazon (Trainium/Inferentia), and Microsoft (Maia) represents a longer-term threat to Nvidia's dominance.

Valuation Considerations

At peak enthusiasm, Nvidia traded at over 40x forward earnings — extraordinary for a semiconductor company. The stock has since moderated, but still commands a significant premium to peers. The bull case requires sustained hyperscaler capex growth, continued Nvidia market share dominance, and successful execution on the Blackwell architecture transition.

Our Assessment

The AI chip supercycle is real, but the rate of growth will inevitably moderate as the initial infrastructure buildout matures. Nvidia remains the best-positioned company in the space, but investors should be prepared for earnings growth to decelerate from the extraordinary pace of 2023-2024.