Summary of Key Points
The current AI industry is shifting from focusing on "what technology can do" to "what practical business applications can achieve." The focus has shifted to how to realize commercial value, the integration of hardware and services, the key elements for enterprises to implement AI effectively, the transformation in investment logic, as well as the advantages and challenges of AI industrialization in China. Forum participants discussed topics such as hardware transformation, the upgrade of application paradigms, factors contributing to success, investment strategies, and opportunities in China, revealing the core principles and future directions of AI commercialization.
Detailed Analysis
1. AI Hardware: From "Functional Addition" to "Service Closure" – A Major Shift in Valuation Logic
In the past, AI hardware (such as smart wearables and humanoid robots) was simply about selling products; for example, a smart watch measured heart rate, and a robot could walk, but these functions did not provide continuous services. The market's response to pure hardware companies has been lukewarm because they merely acted as contract manufacturers without core value.
For instance, a decade ago, a smart bracelet could only count steps, and company valuations were based on manufacturing indicators like "gross margin" and "production capacity," similar to cyclical stocks that were bought and sold quickly. However, if the hardware can collect data, process it, and provide feedback to users (for example, AI glasses that help organize meeting notes or generate PPTs), it creates a "data collection → processing → feedback" service loop. At this point, the company is no longer selling a one-time product but earning money through continuous services. The valuation can soar from a manufacturing multiple of 20 times to a software service multiple of 30-40 times, generating long-term revenue.
2. AI Applications: From "Perception" to "Memory" – The Secret to Doubling Value
Previously, AI applications were limited to the "perception" level (e.g., smart speakers that responded to commands, bracelets that tracked sleep), which merely collected data. Now, they have advanced to the "memory" level: they not only record data but also organize and correlate it to assist in decision-making.
Li Weike's AI glasses are a good example; they can integrate meeting notes and work schedules to quickly generate PPTs (what used to take a week now takes just one day). This "memory capability" transforms the hardware into an "intelligent assistant," and users are willing to pay for the continuous service. The business model has changed from selling hardware with a small profit margin to charging annual fees for the service, resulting in more stable revenue and higher valuations.
3. Implementing AI in Enterprises: Three Critical Elements
PwC surveyed 10,000 CEOs worldwide and found that only 20% of companies have truly made money from AI, with these companies achieving seven times higher cost control and revenue growth compared to their peers. Their success secrets include:
- Top-Level Commitment: The AI strategy must be led by the CEO; otherwise, implementation will fail (e.g., departments working independently without data sharing).
- Data Governance: AI is like a student, and data is the textbook. If the data is disorganized (e.g., inconsistent formats or errors), even advanced models are ineffective.
- Compliance First: Countries have established AI regulations (e.g., the EU's AI Act). If compliance is not considered during design (e.g., data privacy issues), projects may be halted, resulting in wasted investment.
4. Investment Logic: From "Manufacturing Capacity" to "Computing Power Capital"
Microsoft's CEO stated that in the future, there will be two types of core capital for companies: human talent and "Token Capital" (computing power). Computing power is no longer just a cost but an asset like money.
For example, investment research institutions use AI for analysis; what used to take hours can now be done in minutes. However, humans are still crucial because they can make judgments based on experience during market fluctuations when AI provides standardized suggestions.
What are the mature commercialization areas? Code assistance (e.g., AI-driven coding), AI marketing (e.g., automated ad generation), and intelligent customer service (e.g., AI answering customer questions). These areas share common characteristics: standard data (easy for AI training), high demand (companies willing to invest), and clear business models (charging per use or monthly).
5. China's AI Industrialization: Prominent Advantages, but Clear Challenges
Advantages:
- Large Market: A population of 1.4 billion creates numerous use cases (e.g., AI in delivery services and e-commerce).
- Complete Industry Chain: China can produce everything from chips to end products without relying on imports.
- High-Quality Data: Homogeneous language and culture lead to more accurate AI training.
- Rapid Iteration: Companies can quickly adjust their products to keep up with technological trends.
Challenges:
- Global Integration: How can Chinese companies enter overseas markets (e.g., data compliance, user habits)? Hong Kong could serve as a bridge for connecting domestic and global data.
- Data Governance: Although there is plenty of data, much of it needs to be organized and standardized to meet regulatory requirements.
- Human-AI Collaboration: AI can assist with data testing, but final decisions (e.g., investment advice) still require professional judgment to avoid errors.
In One Sentence
The core of AI commercialization lies in the combination of hardware, services, data, and compliance. Those who master these elements will be successful in the future. China has significant advantages, but it must address data management and globalization issues to achieve greater success.