Summary of Key Points
The central argument of this article is that AI is not merely a typical technological bubble or a simple revolution in productivity; rather, it represents the American version of the "urban investment model." This model capitalizes on the narrative surrounding the "future intelligent society," using the capital expenditures of tech giants (such as building data centers and purchasing GPUs) to replace consumer spending by residents, thereby creating a growth mechanism that is detached from the labor and consumption of the general public. This mechanism is more resilient than China's real estate investment model because it does not rely on ordinary households to absorb the costs; it is backed by national strategies and has become embedded within institutional structures (tech oligarchs, capital markets, national security), forming what can be called an "institutionalized bubble." A deeper concern is that AI capitalism might lead to a situation where capital no longer needs humans, but humans could protect themselves through bureaucratic channels and community organizations. In the end, capital and humans may end up abandoning each other.
I. Why is AI referred to as the "American urban investment model"? — A new approach of using corporate spending to replace consumer spending by individuals
The term "American urban investment model" does not refer to local financing platforms similar to those in China; instead, it highlights a similar growth logic:
- Chinese urban investment model: Local governments borrow money to build infrastructure (build roads, create industrial parks), and then rely on developers purchasing land and residents buying homes to repay the debt through land sales and mortgages. The core idea is "future urban value → current infrastructure investment → residents taking over the burden."
- American AI urban investment model: Tech giants invest in building data centers, purchasing GPUs, and securing power supplies, with the expectation that capital markets (stock markets, bonds), national strategies (Sino-US technological competition), and the growing demand for AI within companies will "take over the burden." The core idea is "future intelligent value → current capital expenditure → giants/nations taking over."
The key difference is that, while Chinese urban investment ultimately relies on residents buying homes, American AI urban investment **temporarily does not depend on consumer spending by ordinary people.* Even if you do not use AI services, as long as the capital market believes that "all companies will adopt AI to reduce costs," tech giants will continue to invest. These investments themselves constitute domestic demand (for example, building data centers requires hiring teams and purchasing servers, which generate revenue for other companies).
II. Bookkeeping capitalism: Has production become decoupled from humans? — Capital can function without consumer spending
Traditional capitalism operates on the principle of "humans work → earn wages → consume → companies make profits → reinvest," with humans playing a central role (capital relies on their labor and spending). However, in the era of AI, "bookkeeping capitalism" has changed:
1. First, a "future narrative" is created (e.g., "AGI will transform the world") to drive up the valuation of high-tech companies.
2. High valuations are used to raise funds through bond issuance and stock sales.
3. The raised capital is invested in data centers and GPUs.
4. These expenditures become revenue for the supply chain (for example, chip companies profit from selling GPUs).
5. The profits generated by the supply chain further validate the narrative, driving up valuations.
In this cycle, production remains real (data centers and servers exist), but the connection between humans and production has weakened: capital does not require you to consume AI products first; in fact, it may even see you as a cost factor (for example, companies using AI to cut staff and reduce costs).
III. Why is AI more resilient than Chinese real estate investment? — Four reasons why it is harder to burst
The article argues that AI is more resilient for several reasons:
1. No need for consumer absorption: The collapse of the real estate investment model occurs because no one buys homes, but the collapse of AI would not happen simply because ordinary people stop using AI—data centers remain valuable as long as companies continue to invest in them.
2. Greater flexibility of assets: Real estate values are tied to housing prices and rent, while AI assets (computing power, models, data, cloud contracts) can be valued based on their potential use in the future. As long as there is belief in their usefulness, their value can continue to rise.
3. Fewer players but easier maintenance: Real estate requires the participation of millions of residents, whereas the core players in AI (Microsoft, Google, NVIDIA) are few and can support the narrative that "AI investment is necessary" by serving as both customers and suppliers.
4. National strategic support: AI is at the heart of Sino-US technological competition, so even if returns on investment are low, it will be treated as a strategic asset (e.g., the U.S. government subsidizing chip manufacturers), preventing immediate liquidation.
5. Financial pressure can be delayed: Capital expenditures are recorded and depreciated over time, and debts can be shifted to other entities—short-term profits are not affected, making it harder for the bubble to burst.
IV. Institutionalized bubble: Not just a regular bubble, but part of the system
Regular bubbles (such as the internet bubble in 2000) are characterized by prices deviating from their intrinsic value and bursting suddenly. However, AI is an "institutionalized bubble":
- It has become embedded within the strategies of tech giants, stock market indices, energy systems, and national security narratives (for example, NVIDIA's chips hold a significant share of U.S. equity; AI data centers are key national infrastructure).
- Even if certain parts fail (e.g., a company goes bankrupt), the overall system remains intact—risks can be managed through methods like depreciation, debt extensions, and government subsidies, turning them into "institutional costs" (similar to roads that continue to exist even if they do not generate profit).
In simple terms, AI is not a "false" bubble; it creates real assets through real investments, but the value of these assets depends on narratives about the "future intelligent society," rather than current demand.
V. How should humans adapt in the AI era? — Stop focusing on usefulness and learn to organize
The most concerning aspect of this article is that AI may render capital independent of human needs (labor, consumption, debt). But humans are not machines; what should we do?
- Don't compete with machines in terms of usefulness: AI will eliminate repetitive tasks and even specialized knowledge (e.g., writing copywriting, conducting analysis); you cannot outperform machines in efficiency.
- Organize yourself: Formulate through family, community, unions, and government to transform from being viewed as "tools" by capital into individuals with rights. For example:
- The bureaucratic system provides stable jobs, social security, and channels for grievance redress.
- As a neighbor, parent, or community member, you have identities that prevent you from being reduced to a mere data point.
- Use legal means to advocate for rights such as rest time, compensation for AI-related job losses, and access to public services.
The core logic is that while capital can bypass individuals, it cannot bypass organized human entities. You are not just a labor force; you have identities, rights, and belong to communities.
Conclusion
AI does not aim to allow capital to "ascend freely"; instead, it reveals a problem: when capital no longer needs humans, humans may no longer need capital either. The real risk with AI is not its potential collapse but its transformation into a system that does not require the participation of ordinary people. However, humans can redefine their existence through organizations, rights, and communities, ensuring that the benefits of technological progress do not lead to loss of dignity and security.
In summary, while AI presents challenges, it also offers opportunities for redefining our roles in society. The key is to understand how we can use technology to enhance our dignity and security rather than merely focusing on its utility.