Summary of the Key Points
This article reveals the stark contrast within the robotics/AI hardware sector, particularly in the field of embodied intelligence: on the surface, funding is booming and valuations are soaring (with leading companies valued at over 20 billion yuan), but those working in this industry face immense challenges—heavy capital investment, long periods of trial and error, low tolerance for mistakes, and a reliance on luck to wait for the market to take off. Even experts from traditional hardware backgrounds (such as those from DJI) may struggle to succeed, as robotics represent an “uncertain technical system” rather than a “definite product.” The fatal flaw in AI intelligent hardware is the inability to sell products effectively, while the financing in this area often involves strategic moves aimed at capitalization (pre-packaging companies as assets suitable for the Hong Kong stock market). Ordinary people should not rush into this field; it’s not for everyone.
1. The Realities of Robotics Hardware Professionals
While outsiders talk about a “technological revolution” and a “trillion-dollar market,” those working in robotics deal with mundane and exhausting tasks daily, such as dealing with molds, certifications, inventory management, payment terms, and after-sales services, each of which can be extremely problematic:
- Zero tolerance for hardware errors: Software issues can be fixed with new versions, but hardware defects mean a whole warehouse of defective products (for example, a faulty joint design could result in wasted mold costs).
- Money out first, problems later: Before products are sold, significant expenses have already been incurred for molds, supply chain payments, and certifications. Once products are sold, additional issues arise—high repair rates, pressure from distributors on prices, and customers demanding payment terms, which can quickly lead to cash flow crises.
- Too many uncontrollable factors: Sudden price increases in the supply chain, policy changes, investor withdrawals, or competitive price cuts can all cripple a company. Many companies fail not because of poor products but because they cannot sustain themselves until the market takes off.
These professionals may appear to be bosses or senior employees, but they are actually part of a vulnerable group: bosses are pressured by capital to achieve growth, while senior workers bear the blame despite having little decision-making power. The power and responsibility are inversely distributed—those who set the direction are not held accountable, and those who are responsible have no real authority.
2. Why the Need for Rampant Financing?
The development of embodied intelligence requires funding in three main areas:
1. Robotics hardware: Components like joints, dexterous hands, and sensors are still immature. Suppliers may send 5 out of 10 defective samples, so robotics companies must establish their own supply chains and testing processes, which is costly but has low barriers to entry (since similar components are used by many companies).
2. Data collection: Robots need detailed data on interactions (e.g., the force required to grasp objects or feedback on slipping surfaces), and collecting this data for thousands of scenarios (different bottles, lighting conditions, materials) is extremely expensive.
3. Model development: The technical approach is constantly evolving (today it’s VLA, tomorrow it might be a different model), and each change can render previous data and hardware useless. The cost of computing power and training systems far exceeds that of algorithm engineers.
Financing is not just for research and development but also to secure a strategic position: local governments want to develop this industry, state-owned assets seek investment opportunities, and investors fear missing the next wave of growth. Everyone is betting on the “future,” even though many will fail. The goal is to secure funding while the window is open in order to survive until the next round of market consolidation.
3. Why Traditional Hardware Experts (Like DJI) Struggle with Robotics?
DJI excels with drones because drones are “definite products” with clear goals and technical paths (stable flight, high-quality imaging, specific tasks). However, robotics is an “uncertain system”:
- Lack of standard solutions: Questions such as whether to go for humanoid or wheeled designs, dexterous five-finger hands or two-finger grippers, or whether to target home or industrial applications remain unresolved.
- Mismatch in organizational capabilities: Traditional hardware companies are accustomed to rapid development, fixed BOMs, and mass production, but robotics requires the ability to tolerate long-term trial and error and adapt to changing requirements.
- Difficulty in skill transfer: DJI’s expertise lies in systems engineering and supply chain management, but robotics requires interdisciplinary skills in modeling, data analysis, and scenario assessment, along with the acceptance of uncertainty.
Robots are not just more complex hardware; they represent a completely different approach to understanding and business models. While DJI is good at refining existing solutions, robotics requires finding the right answers from scratch.
4. The Fatal flaw in AI Intelligent Hardware: Inability to Sell Products
Many AI hardware teams focus on showcasing their technology (models, future prospects) but fail to sell products effectively:
- Sales expertise is critical: Selling products means having real customers willing to pay, distributors willing to purchase, and maintaining stable after-sales services. Failing to sell products can lead to high inventory costs, delayed payments, and repair expenses that can bankrupt a company.
- **Avoid the “blue ocean”: What seems like an unexploited market may actually have no real demand. More viable areas are the “red oceans”—existing categories with clear customer needs and distribution channels, such as vacuum cleaners, which just need more intelligence.
- Don’t be arrogant: Don’t blame the market or lack of education if products don’t sell; AI hardware must solve real problems.
5. The Capitalization Strategy Behind Embodied Intelligence Financing
Many embodied intelligence companies raise funds multiple times a year, not just for cash needs but also to prepare for capitalization:
- Boosting valuations: They use small, high-value rounds of financing to establish a valuation foundation and then attract industrial or state-owned capital, raising their valuation to over 20 billion yuan.
- Adjusting the shareholder structure: They bring in local state-owned assets for orders and industry partners to secure supply chains.
- Preparing for Hong Kong stock market listings: They utilize the 18C channel (allowing unprofitable tech companies to list) to package their companies as assets suitable for the secondary market, initially supported by funding and later supplemented by government orders and related ecosystems.
This financing is not about investing in a company but about positioning it as an asset that could be listed in the future, even if it’s not yet profitable.
Final Warning
This field may seem promising, but ordinary people should be cautious:
- Opportunities are reserved for those with advanced technology, industry connections, and capital (e.g., researchers from top universities or large companies).
- Success requires not just effort but also financial resources, resilience, and luck to wait out the market’s challenges.
- Don’t let the hype about a “trillion-dollar market” blind you; most people are just observers, not participants.
Those with poor products or flawed strategies can criticize as they wish, but don’t blame stupidity for failure—many simply fall into a situation with high capital requirements and low tolerance for mistakes.