-
Nvidia's record gains over the last few years can be credited to its flourishing GPU business.
-
Investors may not realize that Nvidia and others outsource much of their foundry services to TSMC.
-
TSMC is poised to continue capturing growth as demand for AI infrastructure scales higher.
-
10 stocks we like better than Taiwan Semiconductor Manufacturing ›
When investors think about artificial intelligence (AI) and the chips powering this technology, one company tends to dominate the conversation: Nvidia (NASDAQ: NVDA). It has become an undisputed barometer for AI adoption, riding the wave with its industry-leading GPUs and the sticky ecosystem of its CUDA software that keep developers in its orbit. Since the launch of ChatGPT about three years ago, Nvidia stock has surged nearly tenfold.
Here's the twist: While Nvidia commands the spotlight today, it may be Taiwan Semiconductor Manufacturing (NYSE: TSM) that holds the real keys to growth as we look toward the next decade. Below, I'll unpack why Taiwan Semi -- or TSMC, as it's often called -- isn't just riding the AI wave, but rather is building the foundation that brings the industry to life.
What makes Taiwan Semi so critical is its role as the backbone of the semiconductor ecosystem. Its foundry operations serve as the lifeblood of the industry, transforming complex chip designs into the physical processors that power myriad generative AI applications.
TSMC manufactures GPUs designed by Nvidia, CPUs for Advanced Micro Devices, and a widening range of custom silicon that cloud hyperscalers are using to optimize AI workloads more efficiently. Today, Taiwan Semi dominates the global foundry market with roughly 68% share of industry revenue -- leaving rivals like Samsung Electronics in a distant second place with just 8%.
One of the louder bear cases against Nvidia and AMD is the growing adoption of application-specific integrated circuits (ASICs). Hyperscalers are becoming highly motivated to design their own silicon -- not only to fine-tune training performance for AI models, but also to reduce reliance on incumbents and push back against their pricing power.
The trend is already visible: Alphabet's Google is rolling out its tensor processing units (TPU), Amazon is deploying its Trainium and Inferentia chips, while Microsoft is experimenting with its own AI accelerators.
For Nvidia and AMD, this shift could translate into slower growth as spending that once flowed directly toward their GPUs is instead redirected to internally developed hardware. For these enterprises, vertical integration isn't just a budgeting exercise; it's a strategic hedge against dominating third-party suppliers.