Summary of the Key Points
This news article analyzes the investment logic in the field of AI hardware through interviews with Li Yannan, the Managing Director of Sequoia China. It covers various aspects such as the initial investment in Yushu Technology, the criteria for evaluating truly AI-powered hardware (active functionality rather than merely adding AI), the business model for AI hardware (generating revenue from the hardware and then using software/community to build a competitive advantage), the need to be cautious about AI bubbles, and the importance of selecting the right entrepreneurs. The article emphasizes that AI hardware is not just about adding AI to existing devices; it's about creating a sustainable commercial system that can withstand market fluctuations.
Detailed Interpretation
1. Investing in Yushu: First, Assess the Founder’s Vision
Sequoia did not invest in Yushu based on the popularity of the robotics industry but was attracted by the vision of its founder, Wang Xingxing. Li Yannan met Wang through a recommendation and was immediately impressed by his idea of using robots to manufacture more robots—this approach aimed to solve the core issue of mass-producing robots efficiently. Early investments are like dating; even if the project faces challenges, if the founder is reliable (like Wang Xingxing, who has a clear vision and perseverance), Sequoia is willing to continue investing. This highlights that the key to early-stage investment lies in trusting the founder's ability and resilience.
2. Truly AI-Powered Hardware: More Than Just “AI + Devices”
Many people think of AI hardware as adding AI features to ordinary devices (e.g., smart speakers with ChatGPT). However, Li Yannan argues that true AI-powered hardware has two essential criteria:
- Essential AI functionality: The device must rely on AI to perform its core functions; otherwise, it’s just a regular product. For example, a smart ring that monitors health and issues alerts would be meaningless without AI.
- Proactive functionality: The device should be capable of taking action on its own, without user intervention (e.g., a fitness tracker that continuously monitors sleep and stops when the phone is turned off, or AI glasses that identify road signs and provide reminders).
Pseudo-AI hardware often seems promising but is unrealistic due to limitations in computing power and battery life.
3. Don’t Rely on “Lossing on Hardware to Profit from Software”
Li Yannan disagrees with the strategy of losing money on hardware to gain revenue through software subscriptions. He suggests that early-stage companies should first make their hardware profitable (even if marginally) before adding software or services. For instance, a smart ring that sells for a certain price and offers health analysis services will attract users who are willing to pay regularly. Pure hardware products can easily fall into price wars, but software and communities (like those of拓竹) can create a competitive advantage, making it harder for users to switch to competitors.
4. Are We in an AI Bubble?
The current high valuations of AI projects may seem crazy, with some teams receiving billions in funding before they’re even ready. Li Yannan believes that the bubble will burst if there’s no significant progress in model development. Great companies must be able to survive economic downturns; financial markets have cycles, and AI trends will eventually cool down. Only those that can endure these challenges will emerge as winners.
5. Choosing Entrepreneurs: Beware of Arrogant and Ill-Informed Ones
Li Yannan is particularly wary of arrogant entrepreneurs who overestimate their capabilities. These individuals often aim for ambitious goals (like “recreating Apple”) but lack the necessary skills and understanding. Early-stage investments should focus on the entrepreneur’s vision and ability to execute it, regardless of their background (whether they come from a large company or are academia graduates). This year, more professors are starting businesses (due to successful cases like DeepSeek), but most are focusing on niche areas (e.g., AI in pharmaceuticals) with limited consumer applications.
In Conclusion
The opportunity in AI hardware lies in creating devices that can perform tasks autonomously and forming a sustainable ecosystem of hardware, software, and community. This will help companies survive market fluctuations. Smartphones won’t disappear but will evolve into central components of multi-terminal systems. Truly AI-powered hardware will take over some of smartphone functions, becoming indispensable tools in our lives that can automate tasks without human intervention.