Summary of Key Points
This speech was delivered by Ma Xiao, the future CEO of Intelligent CEO, based on his five years of experience in AI hardware entrepreneurship. He discussed three main points:
1. The essence of AI hardware is that it serves as the “ears and eyes” for AI to perceive the real world (collecting data from offline scenarios).
2. The reason early AI hardware initiatives failed was that they tried to move too quickly into implementation, encountering issues related to privacy, design, and usability.
3. The current and future direction for AI hardware is to develop “specialized skill hardware” for specific use cases, with the focus shifting towards system-level innovation and ecosystem building.
1. AI Hardware Is Not Just For Show; It’s the “Ears and Eyes” for AI to Understand the Real World
In the past, headphones and cameras were just tools, but today’s AI hardware acts as sensors that connect AI to the real world. For example, AI-enabled headphones can record entire conversations, and AI cameras can capture scenes for analysis—these types of data are essential for large models since the existing data on the internet is becoming scarce. Real-world scenarios (such as a speech or negotiation) contain valuable information that can be lost without hardware to collect it. In short, the role of AI hardware is to transform invisible offline data into something that AI can process and use to better understand the real world.
2. Why Did Early AI Hardware Fail So Quickly? Three Common Mistakes
After the rise of large models in 2022, many AI hardware products (such as headphones connected to ChatGPT or wearable AI devices) disappeared quickly. The problem wasn’t with the product logic itself but with the hasty approach:
- Privacy Issues: Excessive collection of user data without proper privacy protection led to users avoiding using the products.
- Design Issues: Overemphasis on being completely independent (e.g., not requiring a phone connection) resulted in poor performance and battery life issues, affecting the user experience.
- Usability Issues: Uncomfortable designs (e.g., wearable devices) made it difficult for users to use them regularly.
Another common misconception is that entrepreneurs focus too much on a futuristic appearance while neglecting practical user needs. For instance, there are limited options for wearables on wrists or necks, and users often ask, “What value does this device bring to my daily life?” If these questions aren’t addressed effectively, the product won’t sell.
3. Can Only 1% of People Use AI? The Key Lies in “Zero Friction” and Proactive AI Assistance
The AI industry currently faces a stark divide: 1% of users can make full use of AI (e.g., using it as a standalone tool), while 99% use it superficially. To make AI more accessible to the general public, Ma Xiao suggested two solutions:
- Zero-friction data collection: Devices that continuously collect data without requiring user intervention (e.g., always-on headphones that automatically record conversations and decisions).
- Proactive AI services: Providing assistance before users need it (e.g., reminding them of potential pitfalls during negotiations).
In other words, the goal is to reduce the complexity of using AI by automating processes and making it more like a helpful assistant rather than a cumbersome tool.
4. The Right Approach to Developing AI Hardware Today: Customization for Specific Use Cases + Sharing a “Common AI Brain”
Ma Xiao outlined a practical strategy: Customized hardware for specific use cases = hardware with specialized functions. For example, business negotiation headphones could include legal advice features, and children’s cameras could have image recognition and educational capabilities. All these devices would share a central “AI brain” that serves as a common infrastructure for the entire industry. This approach reduces development costs and ensures a consistent user experience across different products.
5. What Will Determine the Success of Future AI Hardware? System-Level Innovation and Ecosystems
After 2026, the competitive advantage in the AI hardware market won’t lie in a single hit product but in system-level innovation. For example, should a device be able to continuously update its features (e.g., adding emotional support functions) and offer a complete service ecosystem (data collection → analysis → proactive assistance). More importantly, it’s about building an ecosystem where various AI devices (headphones, glasses, watches, etc.) can work together seamlessly. Without such services and an integrated ecosystem, products will quickly become obsolete.
In summary, the future of AI hardware lies in providing long-term value through continuous innovation and a comprehensive ecosystem. Instead of focusing on one hit product, the goal is to create an evolving AI assistant that becomes an integral part of users’ lives.
Conclusion
AI hardware acts as the physical extension of AI’s capabilities. The focus should first be on ensuring users are willing and happy to use it, with a user-friendly experience. In the future, success will hinge on building robust systems and ecosystems. For entrepreneurs, it’s essential to consider how their hardware can solve real problems for users. For consumers, future AI hardware will become more intuitive, adapting to their needs without requiring complex setup processes.