NVIDIA is set to reduce shipments to its add-in-card (AIC) partners by approximately 15-20%, as reported by MEGAsizeGPU on X, who has accurately predicted NVIDIA’s product lineups in the past. This supply cut, which is largely affecting GPU shipments, will require AIC manufacturers to adapt to receiving fewer GPU dies than originally planned. Traditionally, companies like AMD and NVIDIA supply raw GPU dies combined with GDDR memory, which AICs then engineer and integrate onto custom-designed PCBs. However, NVIDIA’s decision marks a significant shift in the supply chain, with up to one-fifth of the expected shipments being canceled.
The reasons behind this reduction remain unclear, creating uncertainty regarding its impact on aftermarket pricing and availability. Speculations suggest that NVIDIA may be reallocating its TSMC manufacturing capacity from consumer-grade “Blackwell” chips to server-grade models such as the B200 and B300, but concrete details are lacking.
On a more positive note, NVIDIA continues to bundle GDDR7 memory with the GPUs it supplies to AICs, despite previous rumors suggesting that manufacturers would need to source their own memory from other suppliers like SK Hynix, Samsung, and Micron. This assurance remains a critical factor, as the provision of both GPU and memory kits helps AICs maintain production stability amidst the supply constraints.
Adding to concerns about NVIDIA’s commitment to the consumer GPU market, the company has reportedly halted production of its mid-range GeForce RTX 5070 Ti graphics card, designating it as end-of-life. This decision, attributed to supply shortages, raises questions about the overall focus of NVIDIA on consumer graphics, suggesting a potential strategic shift towards prioritizing higher-end or server-grade products instead. As the industry continues to evolve, NVIDIA’s efforts to adapt to market dynamics will be crucial in navigating these challenges.