Nvidia and Micron have been two of the best artificial intelligence (AI) semiconductor stocks to own, but the easy gains may be over.
Nvidia and Micron have been two of the best artificial intelligence (AI) semiconductor stocks to own, but the easy gains may be over.
As the artificial intelligence (AI) revolution evolves, investors face an increasingly difficult question: Which chipmakers stand to dominate five years from now?
Nvidia (NVDA +1.44%) faces growing competition from custom silicon developed by its own customers. Meanwhile, Micron Technology (MU +4.28%) could face less favorable memory market dynamics as manufacturers expand production.
This is why I see Taiwan Semiconductor Manufacturing (TSM +2.11%), otherwise known as TSMC, as a more attractive long-term opportunity. Taiwan Semi's latest earnings report displayed how its business is becoming more valuable than ever to the AI infrastructure era.
Nvidia and Micron face risks that Taiwan Semi can navigate
Nvidia is the undisputed king in the AI accelerator market thanks to its graphics processing units (GPUs), networking technology, and CUDA software ecosystem. However, the company's largest customers are developing alternatives. Alphabet, Amazon, Microsoft, and Meta Platforms are investing in application-specific integrated circuits (ASICs) optimized for their own AI workloads. While these chips won't replace Nvidia's GPUs entirely, they should reduce dependence on Nvidia while giving the hyperscalers more control over infrastructure costs and model performance.
Meanwhile, Nvidia is also exploring agentic AI, autonomous vehicles, and robotics. Indeed, these opportunities could be enormous. But commercial adoption may take longer than the next four years.
Micron is looking at a different challenge. The company's memory solutions are essential for AI computing, particularly high-bandwidth memory (HBM) and advanced DRAM. Memory manufacturing has historically been a cyclical industry, meaning today's shortages and elevated prices will lead producers to expand capacity. By next decade, more supply should come online, leading to pricing pressure and eroding profit margins.
Moreover, competition from China's ChangXin Memory Technologies (CXMT) also deserves attention. Although CXMT is not yet an equivalent competitor across every advanced memory category, its progress could affect the industry's supply dynamics over time.
Here's what makes TSMC different: Whether AI hyperscalers buy Nvidia GPUs, develop custom silicon, or adopt entirely new architectures, many of those chips will require Taiwan Semi's advanced foundry capabilities. In turn, Taiwan Semi doesn't need Nvidia to defeat Alphabet's custom chip strategy. It can manufacture chips for both companies.
Averytin News Financial update.