Summary of the Core Message
The central argument of this article is that the AI revolution is not an “exception” like the internet; rather, it represents a return to the “normal state” of traditional technological revolutions, such as those in railways and electrification. These revolutions involved investing heavily in infrastructure to overcome hardware limitations before experiencing a surge in applications. Similarly, AI is poised to become a “universal interface for intelligence,” just as electricity has become a universal energy source. This will fundamentally transform how society produces and organizes itself. We are still in the early stages of infrastructure development, and the true giants in the AI industry may not have yet emerged.
Detailed Analysis
#### Why AI is Different from the Internet
The internet was a “scenario revolution”: it moved existing offline services (shopping, chatting, reading news) online, making it easy for users to adopt, and the platform’s marginal cost was almost zero (adding more users hardly increased costs). As a result, applications like Taobao and WeChat became successful quickly, and the platform companies made money first.
AI, on the other hand, is a “productivity revolution”: while the scenarios remain the same (writing articles, providing customer service, programming), the capabilities have significantly improved (for example, AI can write code 10 times faster than humans). However, the existing industrial chains (such as GPU production and data center power supply) cannot keep up with these advancements. Therefore, it is necessary to address these hardware limitations first; those who solve these issues will reap significant profits.
In simple terms, the internet was like “building a road before driving cars” (the road was already partially constructed), while AI is more like “building the road from scratch.”
#### The Normal State of Technological Revolutions: Building the Tools First
Looking back at the past 200 years of technological revolutions (railways, electrification, automobiles), the initial winners were not the end-users but the companies that provided the infrastructure and the necessary tools:
- Railway Revolution: It wasn’t the railway companies that made money; it was steel producers, coal suppliers, and machinery manufacturers (the tools needed to build railways) that profited. Regardless of whether the railways were profitable or not, steel and coal were still required.
- Electrification Revolution: It wasn’t the later-developed appliances (refrigerators, air conditioners) that were successful; it was the power plants, transmission networks, and electrical equipment manufacturers (such as GE) that thrived. People understood the importance of electricity but didn’t know how to use it, so the infrastructure had to be established first.
The internet was an exception because the United States already had a telephone network, fiber optics, and PC servers in place by 1995, allowing the internet to take off quickly on this existing foundation. AI, however, lacks such existing infrastructure, so it must follow the normal path of building the necessary tools first.
#### The Lesson from Electrification: Universal Interfaces are Key
The significance of electricity lies not in its being a new energy source but in its role as a universal interface for energy:
- Electricity can be transmitted, converted (into light, heat, mechanical motion), and controlled (with switches). Steam refrigerators existed before, but only electric refrigerators became popular due to their higher efficiency. Steam factories had to arrange machines around steam engines; with electricity, each machine could have its own motor, enabling the development of assembly lines.
AI is doing the same: it aims to become a universal interface for intelligence, making intellectual tasks (programming, diagnostics, analysis) more accessible and scalable. It turns these complex processes into standardized services (APIs) that can be provided 24/7, allowing small companies to access advanced capabilities just as easily as households use electricity.
#### The Future of AI
Many people think that AI applications will be limited to chatbots and coding tools, but this is just the “primary stage,” similar to how people only knew about light bulbs when electricity first appeared. The real impact of AI will be in transforming how society produces and organizes itself:
- Factories in the steam age were centered around steam engines; with electrification, assembly lines emerged. In the AI era, companies may no longer rely on a large number of highly educated employees but will use AI interfaces to access intellectual services, leading to completely different organizational structures.
AI could even challenge the traditional “factors of production” (laborers, tools, materials). Will AI be considered a tool or a new type of “virtual laborer”? This will redefine work and social structures.
Current AI applications (like ChatGPT) are still at the basic level. The true giants in the AI industry will be those that use it to transform entire sectors—these companies may not have yet emerged.
Conclusion
AI is like electricity in 1880: everyone recognizes its potential to change the world, but we are still focused on building the infrastructure (power plants and transmission networks). The real “game-changers” (applications that will revolutionize industries) are still to come. When investing in AI, don’t rush to look for specific applications; instead, focus on identifying companies that can overcome the infrastructure challenges. This is the true nature of technological revolutions.