Summary of Key Points
The central argument of this article is that the United States and China have "switched roles" in their financing models for the technology industry: China is following the path taken by NASDAQ in 2000, activating an innovation cycle through a large number of tech company IPOs (where companies can go public even without making a profit), while the U.S. is adopting practices similar to China's real estate sector since 2010, using financial leverage such as SPVs, private credit, and asset securitization to support AI infrastructure development. Although these approaches seem opposite on the surface, both involve realizing future demands in advance, they also come with their own set of risks.
Detailed Analysis
1. U.S. AI Infrastructure: Borrowing from China's Real Estate Model with "Shadow Banking"
The U.S. needs substantial funds to build AI infrastructure (data centers, chips), but tech giants are reluctant to dilute their equity (shareholders disagree) and hesitant to take on more debt (rating agencies would downgrade them). As a result, they have copied the "off-balance-sheet financing" strategies used by Chinese real estate developers in the past:
- NVIDIA's New Approach: NVIDIA partnered with six asset management giants to create a financing platform that allows customers to buy GPUs on loan, which are then packaged and sold to pension funds and insurance companies. This essentially turns GPUs into assets that can be mortgaged, rented out, and securitized (Jensen Huang described this as "chips becoming investable for the first time").
- Meta's SPV Strategy: Meta established a shell company (SPV) with financial institutions holding 80% of the shares and Meta only holding 20%. The SPV issued $27.3 billion in bonds to build data centers, allowing Meta to maintain a "low-debt" profile while still needing to rent the data centers for a long term. These bonds received an A+ rating, attracting eager investors.
- Surging Private Credit: Private loans related to AI have increased from nearly zero to $200 billion and are expected to grow by another $800 billion in the next two years. This money represents "shadow banking" (loans not recorded on traditional bank balance sheets), shifting the risks to the institutions and individuals purchasing the bonds.
Comparison with Chinese Real Estate: Similar to how developers used trusts and equity-backed loans to fund construction before selling houses, the U.S. is now realizing the potential revenue from future AI capabilities in advance, transferring the risks to the financial market.
2. Chinese Tech IPOs: Replicating NASDAQ, Going Public Without Profit, Driving Innovation through "Capital Circulation"
China has relaxed its listing policies for tech companies:
- The Fifth Standard of the STAR Market: Large model companies that do not make a profit or have stable revenue can still go public on the A-share market as long as their technology is strong. The Hong Kong Stock Exchange has been doing this for years (for example, allowing unprofitable biotech companies to list).
- The Case of ChangXin Storage: On its first day of trading, its market value surpassed that of Industrial and Commercial Bank of China, making it the largest company in China. Although ChangXin only raised $57.9 billion in cash from its IPO, its market value reached $3 trillion. The key is not the amount of capital raised but the "capital circulation": early investors can exit, employees can liquidate their shares, and the company can use these funds for acquisitions or additional equity offerings.
- Hong Kong Stock Market Financing Boom: Chinese tech companies raised $270 billion in Hong Kong this year, 80% more than last year (with Hong Kong leading global IPOs).
Comparison with NASDAQ in 2000: Back then, U.S. tech companies went public enthusiastically, even those in losses (e.g., VA Linux rose 698% on its first day). The current enthusiasm for new listings in China is similar, with the goal of having the public market take on early risks and inject capital into the industry.
3. The Inevitability of the Role Switch: Driven by Each Country's Specific Needs
Why has this switch occurred? It's all due to practical demands:
- In China: The real estate era is over, and residents' wealth needs new investment opportunities (they can't just rely on buying houses). Tech companies, which previously relied on foreign funds and overseas listings, are now seeking domestic financing channels, with IPOs becoming the best option.
- In the U.S: AI infrastructure is capital-intensive (building data centers and purchasing chips requires significant funding), but tech giants' balance sheets cannot accommodate such debt levels. Shareholders oppose equity dilution, so they have to adopt off-balance-sheet financing methods similar to Chinese real estate developers.
4. The Unique Aspects of Each Model: Differences in Approach and Risks
Despite the role switch, both models have their own characteristics and risks:
- China's Uniqueities:
- The scale of IPO financing is small (a few hundred billion dollars per year, compared to SpaceX's entire funding needs), with bank loans and government industrial funds being the main sources of capital. IPOs in China are more like a showcase than a major source of funding.
- The method of clearing these investments differs from NASDAQ: NASDAQ experienced a collapse (a 78% drop in 2000), while A-share markets have often paused new listings. The outcomes could be different.
- The U.S.'s Uniqueities:
- The risk transmission mechanisms are different: In China, real estate risks affect residential mortgages and banks; in the U.S., AI infrastructure risks lie with private credit and bond investors (although pension funds and insurance companies play a role in absorbing these risks, residents are still affected).
- GPU depreciation is a concern: Houses can last for decades, but GPUs may only be useful for a few years. If depreciation is rapid, the value of collateral could decrease, potentially leading to debt crises.
5. Real Demands, but Avoid Overdrawing from the Future
Whether it's China's tech IPOs or the U.S.'s AI infrastructure, the demand is genuine. Just as railways must be built before passengers can use them, investments in AI and hard technology need to be made in advance. However, potential risks include:
- Reversal of Financing Conditions: If interest rates rise (e.g., with long-term U.S. bond yields hitting new highs), the cost of borrowing will increase, potentially disrupting the entire financing chain.
- Overpricing the Future: For example, if ChangXin's market value reflects expected growth over 10 years, there may be little room for further growth in the future.
However, there's no need to be overly pessimistic: After NASDAQ's collapse, the surviving companies became giants; despite the real estate crisis in China, urbanization was still achieved. As long as the demand is genuine, fluctuations will not prevent value creation—just avoid excessive risk-taking.
Conclusion
Both China and the U.S. are using each other's methods to develop new industries, essentially taking future funds and spending them today. The key is to manage these risks effectively to prevent bubbles from bursting. History repeats itself, but with different actors and rules, the outcomes may vary.