$5,100 across 5 AI chip stocks seen worth $8,900 by 2030

The piece is built on a June Bloomberg Intelligence projection that AI spending across hardware, software, and services will reach $2.3 trillion by 2032, including $658 billion on AI training and $1.3 billion on inference. That forecast frames the semiconductor value chain as a large enough market to produce many winners, and the basket is constructed to cover different parts of that chain: Nvidia for AI training and GPUs, AMD for inference and agentic AI, Broadcom for custom AI chips and networking, Taiwan Semiconductor Manufacturing as the foundry with a virtual monopoly on advanced chip manufacturing, and ASML with its monopoly on extreme ultraviolet lithography.

The valuations are not company guidance but the author's own arithmetic. Each 2030 price target comes from applying a forward P/E multiple to projected earnings per share in 2031, using analyst estimates where available and extending growth assumptions where the consensus does not reach. For example, Nvidia's target applies a 15 times multiple to $24.79 in fiscal 2032 adjusted EPS, while ASML's applies 30 times to $111.72.

What the outcome hinges on is whether those earnings estimates prove accurate and whether multiples hold near today's levels. The author acknowledges this directly, noting that if the AI infrastructure boom goes bust the basket could post a negative return, while if earnings consistently move higher the returns could be larger. For readers, the takeaway is less a forecast than a worked example of how sensitive long-term semiconductor returns are to the earnings path and the multiple the market is willing to pay.