Summary of Key Points
Over the past 20 years, the light-asset model (relying on brands, technology, and platforms to integrate external resources while minimizing the purchase of factories and equipment) has been considered the optimal business strategy. Internet companies have taken this approach to its extreme—such as Meituan not owning restaurants or Didi not owning vehicles. These companies not only earned substantial profits and expanded rapidly but also gained high regard from the capital market. However, with the advent of AI, this model is no longer effective. Large-scale models can now replace the core value of light assets (e.g., SaaS software and traffic entry points), making heavy-asset models more desirable (as exemplified by Goldman Sachs' HALO strategy: a combination of heavy assets and low elimination rates). To survive, internet companies must actively move towards becoming heavier-duty entities—either by investing in their own computing power and data centers or by deeply integrating with the supply chain to have others handle the heavy lifting for them.
I. Why the Light-Asset Model Was Successful in the Past Twenty Years
The essence of the light-asset model is using other people's money and resources to operate one's own business. For example, Nike does not build its own factories but contracts with contract manufacturers; Marriott does not purchase hotels but earns revenue through management services; internet platforms like Meituan and Didi do not need to own vehicles or warehouses, relying instead on algorithms and rules to match supply and demand.
Financially, light-asset companies do not incur significant capital expenditures (Capex) on equipment and building facilities. Their main costs are labor and advertising expenses (Opex), resulting in excellent cash flows. Mature light-asset companies can distribute all their profits to shareholders, while heavy-asset companies can only manage to allocate about half of their earnings. They also expand quickly—opening a franchise is much faster than building a hotel from scratch, and the cost of serving 1 billion versus 10 million users on WeChat is not significantly different.
As a result, the capital market has favored light-asset models: they offer rapid growth, low risk, and attractive financial reports, leading to higher valuations.
II. The Impact of AI on the Light-Asset Model
What was once the advantage of the light-asset model—brands, technology, or traffic—is now being replaced by AI:
- SaaS software is a prime example: Companies like Intuit used to help businesses and individuals with tax filings but have seen their core business disrupted as large-scale models can now handle these tasks directly, causing their stock prices to plummet.
- Consumer-facing internet platforms are also at risk: People used to open multiple apps (WeChat, Taobao, TikTok) to perform tasks, but AI assistants can now do this on their own. In the next 3-5 years, 30%-40% of traffic may shift to AI tools, potentially replacing traditional internet platforms or reducing their role to mere intermediaries with no significant influence.
Light-asset models once held a position at both ends of the value chain (research and development, branding), but AI has now taken over these areas, eroding their competitive advantages.
III. The Rise of the HALO Strategy
The popularity of Goldman Sachs' HALO strategy (a combination of heavy assets and low elimination rates) is not due to the inherent merits of heavy-asset models but rather changes in the times:
- We are living in an era of increasing entropy, characterized by supply chain disruptions, technological challenges, and a reversal of globalization. The traditional model of division of labor (you produce parts, I assemble them) is becoming more problematic.
- The value of heavy assets lies in their resistance to AI: while AI can generate code and write copy, it cannot create factories, data centers, or sophisticated production lines from scratch. These assets require time, money, and experience, creating barriers that AI cannot overcome. For instance, NVIDIA, once a light-asset company, has invested $90 billion in heavy assets to build its own chip-to-data-center-service ecosystem.
IV. Strategies for Internet Companies to Survive AI
To avoid being eliminated by AI, internet companies must either invest heavily or integrate with the supply chain:
1. Investing in Heavy Assets Internally
Companies like Google plan to invest $180 billion in 2026 to build AI servers and data centers, aiming to create a complete ecosystem of models, computing power, entry points, and revenue generation. Meta is also investing heavily, as 98% of its revenue comes from advertising, and AI can enhance ad efficiency; otherwise, it may lose business. Tencent is more cautious but faces criticism for falling behind.
2. Integrating with the Supply Chain
Not all companies have the financial resources to invest heavily, so they opt to partner with suppliers:
- The Apple model: Although Apple does not own factories, it controls the advanced production capabilities of suppliers like Foxconn. By providing technology and setting standards, Apple effectively manages the entire supply chain.
- The WuXi AppTec model: Pharmaceutical companies used to build their own laboratories and research teams (heavy assets), but now WuXi AppTec handles these tasks for them on a project-based basis. This allows pharmaceutical companies to offload their heavy asset burdens, while WuXi AppTec establishes barriers—customers face high costs if they switch to another provider.
In both cases, the goal is to use the strength of assets to ensure resilience against AI.
V. Conclusion
There is no absolute right or wrong between light and heavy assets; it depends on whether a model fits the current circumstances. Light-asset models were successful in the past due to globalization and stable supply chains, but with the rise of AI and increased uncertainty, heavy assets are becoming more crucial. The era of internet-based light assets has ended, but business evolution continues. The next winners will be those companies that invest heavily in key areas and maintain control over their core competencies.
In simple terms, while AI can streamline processes, it is the strength of traditional, heavy-asset models that will provide lasting advantages as the new competitive landscape unfolds.