Nvidia: ChatGPT-Led Boost Drives Our 50% Fair Value Increase

We are raising our fair value estimate on Nvidia stock to $300 per share from $200 per share, as we raise our forecast for Nvidia’s data center segment revenue to grow at a 30% compound annual growth rate over the next five years (up from 19% previously).

AI Drives Demand for Nvidia Chips

The uplift is primarily driven by unsatiated demand for Nvidia’s latest H100 data center GPU.

Relative to its predecessor, the H100 is nine times faster in artificial intelligence training and up to 30 times faster in AI inferencing for transformer-based large language models such as Open AI’s ChatGPT (generative pretrained transformer). We believe AI models such as ChatGPT use thousands of GPUs to be trained, with competition between Microsoft, Google, and others supporting our updated growth projections.

First-quarter sales declined 13% year over year and grew 19% sequentially to $7.2 billion. Gaming sales fell 38% year over year to $2.2 billion as the firm continues to recover from elevated channel inventories. Management believes the channel inventory correction for gaming GPUs is largely in the rearview mirror, which we agree with.

Data center sales grew 14% year over year to $4.3 billion, with strength across cloud service providers, consumer internet, and enterprise customers that are all pursuing various generative AI models. Gross margins rose 130 basis points sequentially to 64.6% thanks to a richer mix led by data center.

Nvidia’s Outlook Boosted by Data Center Demand

We believe the data center segment will be the primary driver of management’s rosy second-quarter revenue outlook of $11 billion. Specifically, we think the robust demand for its H100 GPUs will drive second-quarter data center revenue to $7.6 billion, which would be up 100% year over year.

Furthermore, we think Nvidia will be able to maintain this level of data center sales into the second half of the year thanks to improved supply at TSMC. We now expect Nvidia to double fiscal 2023 data center revenue of $15 billion by fiscal 2025 (calendar 2024) versus our prior estimate of fiscal 2027.

Although we see Nvidia has being the dominant force in AI training and inference for generative AI, the GPU leader’s success is likely to attract intensified competition from AMD, Intel, chip startups, as well as large data center customers themselves that are increasingly designing their own chips to handle AI workloads.