Custom AI Chips Are Coming for Nvidia's Crown. Here Are 2 Companies Quietly Cashing In. - Yahoo Finance

As artificial intelligence continues to advance, NVIDIA, the dominant player in the AI chip market, faces increasing competition from custom-designed AI chips. While NVIDIA remains central to training... Read original →

The surge in custom AI chips is triggering a structural shift across the semiconductor value chain. Technically, domain-specific accelerators like TPUs enforce tighter hardware-software co-design, pressuring EDA, advanced packaging, and interconnect innovations while eroding GPU universality. On compliance, tightening U.S.-EU export controls on advanced computing compel hyperscalers to internalize chip design—boosting supply chain resilience but raising barriers for smaller players. Strategically, NVIDIA retains dominance in model training, yet Broadcom’s ASIC prowess and Marvell’s DPU foothold are capturing high-margin niches; AMD and Intel risk irrelevance without compelling alternatives. Over the next 12–24 months, the market will paradoxically consolidate around closed ecosystems (led by Google/OpenAI) while fragmenting via third-party IP licensing for mid-tier firms. Concentration of advanced-node manufacturing in Taiwan, China and South Korea will exacerbate global supply fragility.