📌 AI Factory Construction Costs Are Squeezing into Hardware Prices
The impact of AI data center investment is spilling over from the server chain into broader hardware prices. What is worth watching now is not whether AI capital expenditure will continue to grow, but that its occupation of wafers, packaging, memory, and related services has already begun to change downstream manufacturers’ costs and quotations.
According to reports, MediaTek and Qualcomm will raise smartphone chip prices in the third quarter, because AI data center construction has pushed up supply-chain costs such as wafers, packaging materials, and services. The core relationship here is very direct: data center customers continue spending money to expand, upstream capacity and materials are squeezed by higher demand, smartphone chipmakers’ procurement costs rise, and part of the pressure is then passed on to the smartphone industry chain. For consumer electronics, this is not a price increase brought by improved demand, but more like a passive transmission of upstream costs.
The transmission on the graphics card side is even more obvious. Nvidia is said to have raised graphics card prices again due to rising memory chip prices, with the increase reaching 20%-30%, and this is also the third price increase this year. This shows that memory is not only a cycle issue for storage companies themselves, but will also affect the end pricing of GPUs and graphics cards. Whoever can pass costs on is more likely to preserve gross margin; whoever is in the middle of the chain and also lacks pricing power will find profits more easily squeezed.
But this line cannot be simply understood as all hardware and storage companies benefiting. There is already a view that the memory boom is approaching an inflection point, and contract chip prices may peak at the end of 2026. Another point that needs to be distinguished is that company earnings do not necessarily change in sync with spot prices. Taking SanDisk as an example, the material mentioned that its multi-year agreements bring USD 42 billion in RPO, which may provide some protection for subsequent revenue. What needs to be verified later is how long price increases can last, and whether contract protection can cover cost and price-down pressure.
▌ Sources