Summary of Key Points
Xiaomi and Alibaba raised funds of HK$42.5 billion and HK$80 billion respectively when they had ample cash on hand. Xiaomi invested in areas such as automobiles and AI-enabled devices, which serve as "physical carriers of intelligence," while Alibaba invested entirely in AI cloud infrastructure. This shift reflects a transition in the digital economy from merely bringing the real world online to integrating intelligence back into reality. For AI to be effective, it requires not only cloud computing power (data centers, chips) but also physical devices (automobiles, robots, smart appliances) – both of which necessitate significant investment in physical infrastructure. By financing in advance, these giants are transforming their balance sheets from mere financial statements into powerful competitive tools.
1. Why Raise Funds When There's No Cash Shortage? – Preparing for the New Pattern of "Spending First, Making Money Later"
Xiaomi will have HK$175.1 billion in cash by the end of 2024, and Alibaba will have HK$474.5 billion by 2026, yet they still choose to issue new shares to raise additional capital. Why? The reason is that investments in AI and smart devices are inherently upfront: for example, building an AI data center requires purchasing chips, constructing data centers, and laying out electrical infrastructure, which only start generating revenue after years when customers start using the services. Similarly, manufacturing automobiles involves setting up factories, developing chips, and establishing supply chains, with returns only coming once sales increase. Waiting until cash runs out to raise funds could result in missing critical technological opportunities or facing a cooling capital market. Raising funds is like stocking up in advance, using a small portion of equity (3.1% for Xiaomi, 3.6% for Alibaba) to secure a steady supply of capital in the long term and avoid potential financial pressures.
2. Cloud Infrastructure: The "Power Station" of the AI Era – with Increased Spending Speeds
In the past, internet cloud services were more like renting out space; adding a webpage didn't incur additional costs. However, generative AI is different. Asking ChatGPT a question or generating an image requires real-time access to models and computing power, which means each use consumes resources. Alibaba's expenditure on capital (for purchasing equipment and building data centers) increased from 3.4% of revenue in 2024 to 12.3% in 2026, and its free cash flow shifted from a surplus of HK$156.2 billion to a deficit of HK$46.6 billion. This is because AI requires more servers, chips, and data centers, which in turn consume more electricity (global AI data center energy consumption is expected to triple by 2030). This infrastructure is not virtual; it consists of physical machines, buildings, and power systems, and the associated costs are much higher than before.
3. Physical Devices: AI Needs a "Body" to Function – Automobiles and Robots as Key Carriers
AI alone cannot make a difference in the real world; it needs a physical form to apply its capabilities. For instance, while AI can plan routes, it needs a car's steering wheel and motors to move; it can detect dirty floors, but it requires a robot's wheels and brushes to clean them. Xiaomi's 2025 capital expenditure is 66% allocated to automobiles, AI, and new businesses, highlighting the importance of physical carriers. Automobiles are an excellent example of intelligent applications, as they combine sensors, computing power, and energy to integrate AI functionality beyond mobile screens.
4. The Digital Economy Returning to the Physical World? – The Inevitable Shift from Virtual to Real
Over the past 20 years, the internet has focused on bringing the real world online (shopping on Taobao, socializing via WeChat). Platforms could profit without owning physical assets. However, AI requires a reversal of this trend: integrating digital intelligence into real devices to enable practical applications. This necessitates two types of infrastructure: cloud computing (the "brain") and physical devices (the "extremities"). These two components complement each other; more devices (e.g., smart cars) generate more data, leading to greater demand for computing power, and stronger cloud capabilities enhance device performance, attracting more users. The digital economy is not moving away from the physical world but becoming more deeply integrated with it.
5. Giant Competition: Balance Sheets as New Fortresses
In the past, internet competition relied on speed (whoever acquired users first and updated software faster won). Now, it's about stability: who can invest in infrastructure in advance, sustain long-term spending, and secure access to essential resources (chips, power). Companies like Microsoft and Amazon in the U.S. spend over $700 billion annually on AI infrastructure. Alibaba focuses on cloud services, while Xiaomi invests in physical devices. Balance sheets (cash on hand, financing capacity) are no longer just financial figures but become competitive tools. Having capital allows companies to build data centers and factories, securing a foothold in the AI era.
In Conclusion
AI is driving the digital economy from virtual services to real-world actions, requiring simultaneous expansion of both cloud and physical infrastructure. Giants are financing ahead of schedule to gain an advantage in this new race between hardware and software.