Hello! I'm your financial analysis assistant. This article from "Huashang Taolue" discusses a very counterintuitive phenomenon: Why do internet giants, which reportedly have billions in cash on their balance sheets, still desperately borrow money from outside sources?
To help you easily understand the logic behind this, I'll first summarize the main points in one sentence and then break it down for you from five different perspectives.
📝 Summary of Key Points
In one sentence:
Although giants like Alibaba, Tencent, and ByteDance have ample cash, they are choosing to raise funds through methods such as issuing shares, bonds, and taking out large loans to lock in capital in advance. This is not just because they are short of money, but also to seize the timing, optimize their financial structure, and take advantage of the currently favorable financing environment in order to prepare for the AI competition ahead.
---
🔍 In-Depth Analysis: Understanding the Logic Behind the Giants' Borrowing
1. Where does the money go? AI is about building infrastructure, not just chatting
Many people still think of AI as something like chatting with robots or generating images, and assume it doesn't require much investment. However, the article reveals a harsh reality: Implementing AI actually involves building substantial infrastructure.
- Visual analogy: The AI services we use may seem lightweight, but behind them are massive data centers in places like Guangling County, Shanxi, with thousands of server racks and specialized power transmission systems.
- Cost of implementation:
- ByteDance: Invested 4.5 billion yuan in the second phase of its Taihang Computing Power Center in Shanxi.
- Alibaba: Announced plans to invest at least 380 billion yuan in AI and cloud infrastructure over the next three years.
- ByteDance's ambition: Internal discussions suggest potential capital expenditures of up to 70 billion US dollars (about 470 billion yuan) by 2026.
- Simple explanation: It's like opening a chain of restaurants. In the past, you might have needed to rent a small space (light assets), but in the AI era, you need to build large central kitchens and logistics warehouses in every city (heavy assets). These data centers are extremely expensive and must be built in advance; otherwise, you won't be able to provide the necessary services.
2. Why borrow now? Time is more valuable than money
If they have cash on hand, why borrow? The reason is that AI competition has a significant delay in infrastructure development:
- Challenges with chips: High-end chips from companies like NVIDIA (e.g., H200) are difficult to obtain and have long delivery times.
- Long construction periods: Building data centers takes a long time from acquiring land to deploying equipment and powering them up.
- Competitive timing: If competitors have already started using advanced hardware, you might fall behind even if you get the money later.
- Simple explanation: It's like buying concert tickets. You might have the money, but the tickets (computing resources/market opportunities) are limited and will expire if you wait too long. Giants borrow now to ensure they have enough resources before the competition starts.
3. How do they borrow? Credit is the key collateral
A notable feature of this financing trend is that loans are unsecured, have low entry barriers, and offer high amounts.
- ByteDance’s example: Borrowed 29.6 billion US dollars (about 200 billion yuan) from nearly 30 banks without any collateral.
- Reason behind this: Banks are willing to lend such large amounts based on ByteDance’s credit and its potential for future earnings.
- Comparison with the past: Two years ago, when ByteDance borrowed 10.8 billion US dollars, the terms were less favorable. Now, with its scale nearly tripling, the terms have improved, indicating that financial institutions still see great value in these leading internet companies.
- Simple explanation: It's like borrowing from friends. In the past, you might have needed to put down a property as collateral, but now, because everyone believes in ByteDance’s future success, friends are willing to lend you a large amount based on trust.
4. Is it a worthwhile investment? Turning cost into profit-generating assets
Previously, buying servers was seen as a pure expense. Now, the logic has changed: Computing power can generate revenue.
- Tencent’s strategy: Executives have mentioned that if their services don’t use all the computing power, they can rent it out to third parties. With the current shortage of computing power, renting it out can almost immediately cover the cost of depreciation and even yield a profit of over 30%.
- Alibaba’s strategy: Alibaba’s CEO, Jack Ma, stated that AI investments will pay off in about three years, possibly even in two. Moreover, older servers (e.g., V100 from 2018) are still in full use, outperforming expectations in terms of lifespan.
- Revenue evidence: Alibaba’s AI revenue has been growing for 12 consecutive quarters, with an annualized figure of over 49.5 billion yuan; Kuaishou’s Keliing AI service generated over 850 million yuan in a single quarter.
- Simple explanation: Buying servers used to be like buying a printer, a consumable. Now, it’s like buying a taxi: you can use it for your own needs and also rent it out for extra income. Since the investment makes sense and can even be profitable, giants are willing to borrow to expand.
5. Why not use all their own money? Financial wisdom and planning for the future
Finally, why not just spend the 400 billion yuan in cash on hand? There are two main reasons:
- Financial structure optimization: The company needs the cash for various purposes, such as buying back shares, acquiring other companies, and dealing with unexpected risks.
- Infrastructure is a long-term asset that should be financed with long-term funds (e.g., 30-year bonds or loans), which is the best financial practice.
- Simple explanation: It’s like buying a house: You should use a mortgage (long-term debt) rather than spending all your emergency savings (short-term liquidity) at once. This way, you stay flexible and prepared for any unexpected situations.
- Seizing the financing opportunity: The capital market is currently enthusiastic about AI, making it easier to obtain funds.
- In the next few years, all tech companies (in areas like chips, cloud, and large models) will need to spend heavily, and funds may become scarce, increasing financing costs.
- Simple explanation: Borrow now while banks are willing to offer low interest rates. If you wait until everyone is competing for loans, interest rates will rise, and banks might become more selective.
💡 Conclusion
This wave of financing may seem like a need for money, but it’s actually about strategic positioning.
In the AI era, technology determines whether you can participate in the competition, and funds and infrastructure determine how long you can stay in the game. By locking in capital in advance, companies like Alibaba, Tencent, and ByteDance are not just buying servers; they are also buying time and certainty. For everyone, it’s important to understand this: Future tech competition will not be about lightweight innovation but about the endurance of heavy infrastructure.