People have been asking me lately whether it’s still possible to buy AI U.S. stocks. Honestly, instead of asking whether you can buy, it’s better to first figure out which part of the industry chain you want to buy into.

AI is not a single industry—it’s an entire supply chain, and the logic across upstream, midstream, and downstream is completely different. Upstream is computing power hardware (like NVIDIA and TSMC). Midstream is platforms and cloud services (Microsoft, Amazon, Google). Downstream is application software (Salesforce, Adobe). Each layer earns money differently, and the stock price drivers are also very different.

I recently looked at some data: the four major cloud giants’ combined capital expenditures in 2026 are approaching $600–700 billion, which directly shows one thing—demand for AI chips is nowhere near fully met. NVIDIA currently holds about 80% to 90% of the revenue share in the AI accelerator market. Each year, it generates more than $100 billion in revenue just from data center GPUs. That figure is staggering, but it also reflects how hungry the market is for computing power.

TSMC is also worth paying attention to. Nearly all of the AI chips from NVIDIA, Apple, and AMD are produced by TSMC. In the first quarter of 2026, combined revenue increased 35% year over year. High-performance computing accounted for as much as 58% of revenue, with a year-over-year growth rate of 48%. Most importantly, since the beginning of this year, TSMC has launched a continuous, four-year price increase for all advanced processes below 5nm. The customers know they’ll face price hikes for four years, yet they still rush to place orders. What does that indicate? Pricing power in the supply chain is gradually shifting.

As for the midstream, Microsoft has successfully introduced AI technology into global enterprise workflows through its exclusive partnership with OpenAI, integrating it via the Azure AI cloud platform and Copilot. Amazon, meanwhile, forms a complete closed loop by combining AWS with its self-developed AI chip, Trainium. Meta took a different route—by optimizing ad AI and using the open-source Llama model, it monetizes AI directly. As AI greatly improves targeting accuracy, the ad placement precision for Facebook and Instagram has improved significantly.

But there’s one problem I’ve been thinking about: what happens when midstream costs are continuously pushed up as upstream chip prices keep rising? NVIDIA’s gross margin is as high as 75%, and some cloud customers have already started developing their own chips to reduce costs. In the long run, this could affect the profit structure across the entire supply chain.

If you ask me how to participate in AI U.S. stocks, my recommendation is to approach it in layers. If you want lower volatility, you can choose Microsoft, Amazon, and TSMC—these companies have solid fundamentals, and AI is only one of their growth drivers. If you want to capture mainstream capital, you can look at NVIDIA and Meta, but you need to be prepared to withstand volatility. If you want to be more aggressive, you can consider second-tier AI chip companies and AI application startups, but that also comes with the highest risks.

To be frank, AI concept stocks have accumulated huge gains over the past two years, and many companies’ stock prices have already priced in years of growth expectations. Once growth slows down or market sentiment turns, the pullback could be quite substantial. That’s why basic skills like building positions in batches, waiting for dips, and controlling a single-stock position size still need to be done.

In the long run, I believe AI’s impact on human life will be no less than the internet revolution back then. But in the short term, there may still be fluctuations, because capital flows are easily affected by factors such as interest-rate policy and the emergence of new themes. Rather than blindly chasing highs, it’s better to keep focusing on a few key factors: whether the development speed of AI technology has started to slow down, whether the monetization ability of related applications is improving as expected, and whether the profit growth rate of individual companies is showing signs of deceleration. As long as these conditions remain, the investment value of AI U.S. stocks is still there.