Summary of Key Points
Recently, there has been a surge in mega-IPOs by AI-related companies around the world (such as SpaceX, ChangXin Technology, and Yushu Technology). These companies generally rely on substantial capital expenditures to sustain their growth—either by cutting back on research and development to squeeze out profits or by expanding their spending after mergers and acquisitions. IPOs have become a necessary option for them to “stay afloat.” The capital expenditure across the entire AI industry has surpassed even the peak of historical bubbles (such as the rail and internet bubbles), raising concerns about debt risks. Behind this trend lies a hidden transfer of wealth: early investors and corporate executives are cashing out, and ordinary investors who buy in at high prices may end up bearing the brunt of any bubble burst.
I. Mega-IPOs: The “Life-Sustaining” Option for Cash-Strapped Companies
On the surface, some of these AI companies that are rushing to go public appear profitable, but in reality, they are all relying on money to survive:
- ChangXin Technology: Despite earning 5.3 billion yuan in 2025 (with a gross margin of 41%), it still needs to raise 29.5 billion yuan in funds because storage chip technology evolves rapidly, and without investing, it would be outcompeted.
- Yushu Technology: Its R&D expense ratio was previously less than 8% (one-third of its peers’), but it managed to squeeze out a profit of 600 million yuan by cutting back on R&D. During its IPO, it plans to invest 85% of the raised funds in R&D, essentially admitting that its previous profits were achieved through cost-cutting.
- Zhipu: It earned 724 million yuan in 2025 but spent 3.18 billion yuan on R&D (4.4 times its revenue), resulting in a loss of 470 million yuan. AI large models are extremely costly, and without continuous funding, these companies cannot operate.
- SpaceX: After merging with Musk’s xAI project, it spends 13 billion yuan annually on chips and data centers, a pace that has worried analysts, making the IPO its only source of funding.
In essence, the AI industry is a “arms race” where those who don’t invest enough will fall behind. An IPO is not an optional choice but a necessity.
II. The “Great Leap Forward” in Capital Expenditure Across the AI Industry
The explosion in demand for AI has led to unprecedented spending across the entire supply chain:
- Demand for computing power: China’s daily AI token consumption increased from 100 billion at the beginning of 2024 to 180 trillion in February 2026, with an expected further increase by 370 times. Each AI interaction or image generation requires tens of thousands of matrix calculations, meaning the demand for computing power is 10,000 times that of traditional internet applications.
- Cloud providers investing heavily: The four major players (Google, Microsoft, etc.) plan to spend a total of 660 billion yuan in 2026 (far exceeding previous forecasts for the top eight global companies). Alibaba Cloud will invest 380 billion yuan over three years, and ByteDance plans to invest 160 billion yuan in 2026—all of this is going towards purchasing NVIDIA chips and building data centers.
- Capacity shortages: Expanding storage chip production requires clean rooms and EUV lithography machines. TSMC’s HBM (high-bandwidth memory) capacity is also limited, leading to a supply-demand imbalance that will persist for 2–3 years.
- Expenditure exceeding revenue: In 2025, the US AI industry’s capital expenditure was six times its revenue (compared to twice during the rail bubble and four times during the internet bubble). Companies are relying entirely on external financing to cover their costs.
III. Rising Debt Levels: A Sign of Crisis?
To fund their spending, companies are borrowing heavily, accumulating risks:
- Surging bond issuance: In 2025, global tech companies issued 428.3 billion yuan in bonds. The five major AI companies on the US stock market (Microsoft, Google, etc.) issued five times more bonds than the average from 2020 to 2024. Chinese companies like Kuaishou, Tencent, and Alibaba are also accelerating their bond issuance (e.g., Tencent restarted domestic financing after a four-year hiatus, and Alibaba issued interest-free convertible bonds).
- Hidden risk instruments: CLOs (Collateralized Loan Obligations) are becoming popular again. These packages high-risk AI loans into different risk levels for investors, meaning that if an AI company encounters problems, the risks could spread like during the subprime mortgage crisis, affecting ordinary investors as well.
- Valuation bubbles: The price-earnings ratios in the AI sector have exceeded 100 times (far exceeding the internet bubble of 2000). Nine out of the top 10 highest-performing stocks on the US stock market are AI-related, and 60% of the double-digit gainers in the A-share market are also AI-related. The bubble is growing larger and could burst at any time.
IV. Wealth Transfer: Who Is Cashing Out, and Who Is Buying In?
Behind this IPO trend lies a silent transfer of wealth:
- Early investors and executives leaving the scene: Leading firms like Sequoia Capital, SoftBank, and Bridgewater have started reducing their holdings in AI companies. Zhu Yiming, the controlling shareholder of GigaDevice Technology, cashed out 2.5 billion yuan in 11 consecutive days before ChangXin’s IPO.
- **The luxury housing market in cities like Shenzhen and Hangzhou is filled with new wealthy individuals from the AI and semiconductor industries (who have cashed out).
- Risks for ordinary investors: Funds in the AI sector are highly concentrated (the seven major US companies account for 40% of the index weight, and TMT transactions in the A-share market account for 40%). If the bubble bursts, ordinary investors who bought in at high prices will suffer significant losses.
- Ray Dalio’s warning: “The process of bursting a bubble is also the process of converting wealth into cash.” Early investors are converting their stocks into cash, while those who buy in later will end up with depreciating assets.
A Reminder for Ordinary Investors
Don’t just focus on the “revolutionary stories” surrounding AI. Understand the true nature of this financial game: companies rely on IPOs to stay afloat, and capital seeks to profit from the bubble. If ordinary investors blindly follow the trend, they may become the ultimate victims. Instead of chasing high-priced AI stocks, it’s better to understand your role in this game—whether you are a “cash-outter” or a “buyer.”
(The entire text is written in plain language to make financial and business concepts accessible to non-experts, explaining the risks and wealth dynamics behind these mega-IPOs.)