Summary of Key Points
The global wave of AI investment continues to accelerate, with technology giants (from both China and the United States) not only showing no signs of slowing down despite the debate over an "AI bubble" but also increasing their capital expenditures (funding long-term investments such as purchasing equipment and building data centers). The scope of investment has expanded from individual components like GPUs to the entire value chain, including storage, data centers, and power infrastructure. However, the resulting cash flow pressures have forced these giants to rely on external financing sources (such as issuing bonds and increasing stock offerings). The focus of the market has shifted from whether to invest in AI to whether the substantial amounts of money being invested will generate profits.
Detailed Analysis
1. Despite the Debate over the AI Bubble, Giants Are Still Spending Heavily
Many argue that AI is a bubble this year, but the actual investment activities of the giants indicate that they have not stopped. For example, the four major cloud providers in the U.S. (Amazon, Google, Microsoft, and Meta) plan to spend $730 billion on AI-related investments in 2026, a 78% increase from the actual expenditure in 2025. Even more strikingly, Meta has revised its investment plans three times within a year, and Google and Amazon have also continued to increase their investments. This suggests that the giants believe the demand for AI is still on the rise, and the investment cycle is far from peaking. Instead of debating whether it's a bubble, it's more practical to acknowledge that this round of AI infrastructure development is still in its expansion phase.
2. AI Investment Has Shifted from Buying Chips to Building Heavy Asset Empires
In the past, AI investments were primarily focused on acquiring GPU chips. Now, the money is flowing into a broader range of infrastructure, such as high-performance storage (HBM, DDR5), which accounted for 8% of AI data center costs in 2023 but is expected to rise to 30% by 2026. The power demand for data centers is also increasing significantly; in the U.S., the electricity consumption of data centers is expected to grow from 50 GW in 2024 (equivalent to the power output of 50 large nuclear power plants) to 134 GW by 2030. In other words, AI investment has evolved from purchasing individual components to building large-scale infrastructure, turning into a competition to invest in heavy assets.
3. AI Investment in China and the U.S.: Similar Growth Rates, but a 11-Time Difference in Scale
Chinese tech companies are not lagging behind. ByteDance, Alibaba, Tencent, and Baidu plan to invest a total of 460 billion yuan (about $64 billion) in 2026, a nearly 40% increase year-over-year, on par with the growth rates of their American counterparts. However, there is a significant difference in scale: the $730 billion invested by the four major U.S. companies is 11 times that of the Chinese companies. This implies both challenges (the U.S. has a more solid foundation) and opportunities (China still has significant room for expansion) for its domestic industry chain.
4. Running Out of Funds? Giants Are Seeking External Financing
In the past, giants could afford to fund their AI investments using their own cash flows. However, with the current scale of investment, they are running out of money. The cash flows of U.S. giants are under pressure; for example, Meta and Google are facing financial challenges, and they have started to issue bonds, with debt accounting for 32% of their investments in 2025 (up from 9%). Chinese companies are also taking action: Alibaba has issued 710 million additional shares to raise HK$80 billion for AI investments, and Tencent's cash flow turned negative for the first time in the second quarter of 2026 (earnings were not enough to cover expenses). Although Chinese companies still have substantial cash reserves, their ability to secure financing has become a critical factor in their ability to continue investing in AI.
5. The Biggest Question: Can All This Investment Be Profitable?
The main question for the market is no longer whether to invest in AI but whether the invested money will actually generate profits. Can these trillion-dollar investments be converted into customer orders? Can these orders lead to revenue? And can that revenue result in profits? For instance, who will win the contracts to build data centers or supply storage solutions for these giants? Only these outcomes will truly determine the success of these investments. After all, the key to determining whether it's a bubble or not lies in whether there are real returns.
Conclusion
AI investment is still expanding rapidly, but the approach has changed. It has shifted from a focus on lightweight components (chips) to heavy infrastructure, and from relying on internal cash flows to seeking external financing. For the general public, the debate over the bubble is less relevant than identifying which companies will actually benefit from these investments. After all, the real value lies in the tangible profits generated by these investments.