Summary of Key Points
This article discusses personal survival and asset allocation in the era of AI, using a paper automatically generated by the Federal Reserve as an entry point to reveal two key phenomena. Firstly, the high valuation of AI stocks is essentially a form of “doomsday insurance” – people use them to hedge against the risk of being replaced by AI. Secondly, ordinary individuals who simply use AI to improve efficiency fall into a trap where their value decreases as they become more skilled; meanwhile, the core AI capital (such as OpenAI and Anthropic) has already been acquired by top-tier investors, making it inaccessible to the general public. The article proposes a solution: transform the efficiency brought by AI into assets that you can own, such as brands, channels, or products, rather than merely using AI as a tool.
1. AI Stocks: Why Are They Considered “Doomsday Insurance”?
The Federal Reserve’s paper, “Hedging the Singularity,” provides an counterintuitive explanation for the high valuation of AI stocks. It suggests that the high valuations are not solely due to optimism about AI companies’ profitability but also because they serve as a form of insurance against the risk of being replaced by AI in the future.
For example, if you are a designer and your core asset is your design expertise, which will generate income for decades, the sudden emergence of AI that can create top-quality designs could drastically reduce your earnings. However, the stocks of AI companies that develop such technologies would soar. In this case, owning AI stocks would provide insurance against the risk of losing your income.
The paper cites compelling figures: even if the probability of AI replacing you is only 1% per year, the valuation of AI stocks would still be twice that of non-AI stocks; if the probability drops to 0.5%, it remains 40% higher. This indicates that the high valuations in the AI sector are not a bubble but rather a collective “risk premium” paid by the global middle class.
However, the reality is harsh. The AI stocks you can buy (such as those of NVIDIA and Google) are merely peripheral assets. The truly core AI companies (like OpenAI and Anthropic) have not gone public yet, and their equity has been acquired by sovereign funds and top venture capital firms. What you think you’re buying is a piece of the future AI revolution, but in reality, you’re only getting a small portion of the benefits.
2. The Trap of Using AI to Improve Efficiency: Are You Worth Less as You Become More Skilled?
Many believe that those who use AI will outperform those who don’t, but this statement obscures a crucial fact: the gap in tool usage will quickly narrow, and relying solely on AI to improve efficiency can make you less valuable.
There are three reasons for this:
1. Tool Simplification: What used to require complex instructions (e.g., writing prompts) can now be done by large models that understand your intentions. In two years, the need for such prompts may become as obsolete as dial-up internet access.
2. Ability Encapsulation: Similar to how you don’t need to understand chips when using a smartphone, in the future, you won’t need to know about model fine-tuning; platforms will handle the complex operations behind the scenes.
3. Rapid Replication of Usage: Any unique AI skill can quickly become a paid service or a standard practice within companies.
As a result, when everyone can use AI for tasks like writing, drawing, or coding, the market value of these skills will plummet. According to Nobel laureate economist Daron Acemoglu’s models, automation shifts income from the labor side to the capital side. The more you rely on AI to improve efficiency, the more your job is at risk of being devalued – a gradual process akin to being boiled in warm water.
3. The Core AI Benefits: Out of Reach for Ordinary People?
The most substantial benefits of AI are not in the public market but in unlisted core companies. For example, OpenAI could be valued at $500 billion by 2025, and Anthropic could be worth trillions by 2026; both are still private. Why don’t these companies go public? The private market offers ample funding (from sovereign funds and family offices), and their rapid spending is beyond the capacity of the public markets. Additionally, they are strategically invested by tech giants like Microsoft and Amazon.
A report from Silicon Valley venture capital firm a16z indicates that the profit margins in the generative AI industry are largely held by companies in the infrastructure (chips, cloud services) and closed-source models layers, while the application layers (e.g., AI writing and design tools) have thin profit margins. The deeper you delve into using AI, the closer you get to the bottom of the value chain.
4. Three Types of People in the AI Era: Which Side Are You On?
The AI revolution will naturally divide people into three categories:
1. AI Capital Owners: Those who hold equity in model companies, chip companies, or possess computing power and data assets. The stronger AI becomes, the more they profit – they can simply reap the benefits without much effort.
2. AI Leveragers: Those who use AI to enhance their brands, products, or channels. For example, using AI to run a personal studio and sell courses or templates turns AI into a source of unlicensed leverage.
3. AI Price Bearers: Those who use AI to increase company efficiency but lose bargaining power. They work for companies, helping them save costs, but their value decreases as a result.
Former Greek Finance Minister Varoufakis aptly describes this as “digital feudalism,” where platforms and AI combine to reinforce the owners’ control over users.
5. How Ordinary People Can Break This Cycle: Turning AI Efficiency into Assets
Since you can’t acquire core AI capital, ordinary people should focus on converting the value created by AI into assets they can own. The following six types of assets are particularly important:
1. Brand Assets: While AI can help with content creation, trust is built through personal expertise, not just AI-generated text.
2. Channel Assets: Your own social media platforms, communities, and email lists create direct connections to readers that are less susceptible to algorithm changes.
3. Product Assets: Encapsulate your experience in courses, templates, or paid reports.
4. Community Assets: Gather high-quality people around a common interest (e.g., an AI enthusiast community), as the value lies in the network and reputation you build.
5. Data Assets: Collecting unique data in specific fields; once general-purpose models become cheaper, your exclusive data becomes competitive.
6. Workflow Assets: Automate your work processes (research, writing, sales) to create assets that serve your own brand.
Naval Ravikant has said that “unlicensed leverage” (such as code and media) doesn’t require approval from others. AI makes such leverage more accessible to everyone. If you can use AI to do what used to take ten people’s work, you should turn the results into assets rather than working for others.
In Conclusion
The dividing line in the AI era is not about whether you can use AI but whether you can convert its efficiency into your own assets. The truly dangerous situation is being among those who use the most advanced AI every day without any tangible benefits. You have the choice: to be the master of AI or just another tool in its hands.