Summary of Key Points
There has been a noticeable shift in AI investment in the U.S. stock market during the second quarter: funds have moved away from companies like Nvidia, which focus solely on providing computing power, to broader AI infrastructure providers such as Micron, Intel, and AMD. The combined market value of these companies has increased by $2 trillion, but sector valuations and volatility have reached historic extremes. At the same time, AI investment has entered a new phase, with the bottleneck shifting from a shortage of hardware to the feasibility of financing computing power, making the efficiency of capital utilization a key competitive factor.
1. AI Investment Shifts: From Nvidia's Dominance to a More Diversified Landscape
Previously, AI investments were primarily focused on Nvidia, the leader in GPU chips, which provided the necessary computing power for AI training. However, the second quarter saw a change in focus. Investors realized that AI requires not only GPUs but also memory, CPUs for general computing, and network equipment as part of the infrastructure. For example, Micron's stock price rose by 240% in the second quarter, Intel's by 216%, and AMD's by 186%, with their combined market value increasing by $2 trillion, while Nvidia only saw a 15% increase, becoming the underperformer in the semiconductor index. The Philadelphia Semiconductor Index (SOX) soared by 81% during the quarter, driven by these companies across the entire supply chain.
Analysts at Barclays believe this marks the beginning of a new phase in AI investment: from focusing solely on the most critical computing chips to investing in the entire infrastructure ecosystem that enables AI applications.
2. The Profit-Making Strategies of the Rising Stars
Why have these companies seen such strong performance? Each has its unique advantage:
- Micron: It cut off its consumer business (such as mobile memory) after 29 years and focused on producing memory for AI data centers. AI chip manufacturers are eager to purchase its memory, driving prices up significantly. Its revenue increased by four times in the latest quarter, with gross profit margins rising from 39% to 84.9%. Micron is currently only meeting 50%-75% of customer demand and intentionally controls new factory construction to avoid oversupply and price erosion.
- Intel: As AI expands into devices like smartphones and computers that require CPUs, Intel's CPU business has seen a surge in demand, leading to significantly higher-than-expected revenue in the second quarter.
- AMD: Although its GPU performance lags behind Nvidia, its CPU and data center businesses are strong. Its revenue exceeded expectations in the first quarter, and its data center business prospects for the second quarter are also promising, driving its stock price up.
Other companies, such as Mellanox (network equipment) and Arm (chip design technology), have seen significant gains due to the expanding demand for AI infrastructure.
3. Concerns Behind the Surge: Valuation Bubbles and Increased Volatility
While the rapid growth is impressive, it also raises concerns:
- Excessively High Valuations: The current forward P/E ratio of the SOX index (stock price divided by expected future earnings) is 26 times, higher than the average of 19 over the past decade. Intel's P/E ratio is at 100 times, meaning it would take 100 years to recoup the investment based on current earnings, while Arm’s ratio is even more exaggerated at 140 times. In contrast, Nvidia’s P/E ratio is only 18 times, making it relatively affordable.
- High Volatility: The volatility index for semiconductor ETFs has increased by 83% this year, the largest increase on record. The SOX index fluctuates by more than 1% daily, with some days seeing sharp gains or losses. Last week, semiconductor ETFs fell by 7.3%, the worst performance in nearly a decade. Institutions are concerned that the current market focus on AI infrastructure may not be sustainable.
- Retail Investors Still Participating: Micron was the most heavily bought stock by retail investors this Monday, with over $100 million invested in two days. However, institutions warn that if profit expectations are too optimistic, prices could quickly correct.
4. The New Phase of AI Investment: From Hardware Competition to Financing and Efficiency
Companies like Lemborgh are pointing out that the bottleneck in the AI industry has shifted. No longer is there a shortage of GPUs; the challenge is now in financing the construction of data centers that require substantial resources such as electricity, land, and cooling systems. Additionally, companies must manage their debt levels to avoid financial strain.
Large technology firms, which were once characterized by light assets (high profits and cash reserves), are now investing heavily in data centers for AI infrastructure, disrupting traditional business models. In the future, those who can use capital most efficiently to build the most computing power will emerge as winners. Sovereign investors (such as national funds) are also interested in AI, but transforming their interests into investments requires a broader perspective, covering software, physical infrastructure, energy systems (which require significant amounts of electricity), and industrial capabilities, rather than simply buying individual stocks.
Conclusion
AI investment is no longer dominated by Nvidia alone; the entire supply chain is benefiting. However, investors must be cautious of valuation bubbles and increased volatility. The future of AI will depend not only on hardware but also on the ability to fund and operate data centers efficiently. For individual investors seeking to participate, it is important to assess whether company performance matches their expectations and valuations before making decisions.