I have been closely monitoring the performance of U.S. AI concept stocks lately and found that this AI wave is truly reshaping the entire tech investment landscape. From the launch of ChatGPT to now—about two to three years—the price gains of the related names have been nothing short of extraordinary.

When it comes to AI investing, the chip supply chain is definitely impossible to ignore. NVIDIA can be said to be the biggest beneficiary—this company’s GPUs have become the industry standard for training and running large AI models. From chips and systems to software, its complete ecosystem makes it the core of AI infrastructure. In its recent financial results, its net profit growth rate exceeded 200%, which is really uncommon among tech stocks. What’s interesting is that even TESLA’s boss and Oracle’s founder have privately asked NVIDIA’s CEO for chips, showing just how strong the market’s demand for computing power is.

However, although NVIDIA is strong, AI development cannot happen without the entire industry chain. Broadcom is also worth paying attention to, as it has clear advantages in network chips and data center interconnects. AI servers require high-speed networking support, and Broadcom’s ASIC chips, network switches, and optical communication chips are all necessary. Interestingly, although NVIDIA and Broadcom compete in some areas, they are actually complementary—both have been rising during this AI boom.

As NVIDIA’s direct competitor, AMD has also been stepping up its efforts over the past two years. Its MI300 series accelerators have matched NVIDIA’s H100 in many tests; the key advantage is that they cost only half as much as the H100. This is especially attractive to large enterprises that want multiple-source supply. Although NVIDIA’s CUDA ecosystem gives it a temporary lead, as AI applications become increasingly diversified, AMD’s opportunities are also growing.

In terms of market size, IDC predicts that in 2025, global enterprise spending on AI solutions will reach more than $300 billion, and by 2028 it may break $600 billion. The growth room is indeed substantial, and infrastructure-level investment makes up the majority—good news for chip and server manufacturers.

There are actually quite flexible ways to invest in U.S. AI concept stocks. In addition to directly buying individual stocks, you can also diversify risk through thematic ETFs. For example, some AI and big data funds already have global assets of more than $30 billion. If you’re doing short-term trading, you can also consider trading via a contract for difference platform, which allows more flexible position adjustments.

That said, it’s important to note that risks in this space are not small either. AI technology is developing quickly, policies and regulations could change at any time, and valuations themselves are also relatively high. For some companies, their stock prices may have already priced in the future positive developments. So when it comes to investment strategy, I think entering in batches and holding long-term is more reliable than chasing highs in the short term. After all, this AI wave is only just beginning, but we should also be alert to the possibility that at some point the market’s focus may shift to other sectors.

Overall, U.S. AI concept stocks still offer opportunities in infrastructure areas such as chips, cloud services, and data centers. But the key is to find truly competitive companies, not to blindly follow the trend. The market is always changing, and investments need to be adjusted accordingly.