Summary of Key Points
Alibaba recently raised HK$80 billion (approximately $10.2 billion) by issuing new shares, all of which will be invested in the AI sector. Its financial report shows a 57% year-on-year decrease in profits, but capital expenditures on AI infrastructure (such as purchasing equipment and building data centers) increased by 75%. This trend reflects the global tech giants' frantic spending on AI computing power, as there is a consensus that computing power will be in short supply by 2030, and investments can yield quick returns (within 3 years or less). The AI race has shifted from competing on models to competing on infrastructure, with the ultimate outcome depending on a combination of capital, computing power, and energy capabilities. Chinese companies that do not keep up may lose their leading positions.
Detailed Analysis
1. Why is Alibaba willing to invest so heavily in AI?
This is Alibaba's first time raising funds by issuing new shares since its listing on the Hong Kong stock market in 2019, and all the money is being invested in AI. Why such a bold move?
- Short-term sacrifices for long-term opportunities: Although selling new shares will dilute the shares of existing shareholders, AI computing power is the new "infrastructure" of the future. Just as companies competed for internet traffic 20 years ago, they are now competing for computing power to gain control of the future.
- The logic of investing despite declining profits: Although profits have decreased by 57%, capital expenditures (CAPEX) have increased by 75%, with most of this going towards AI data centers. Alibaba CEO Daniel Wu emphasizes that AI is a "capital-intensive industry" that requires upfront investments in GPU chips and data centers to provide AI services (such as large models and intelligent customer service). Failing to make these investments early will result in missing out on market opportunities.
- Collaboration to share the burden: Alibaba is not working alone; it partners with other companies to build computing power centers and even has customers pay in advance to reduce cash flow pressures. For example, its collaboration with Envision Energy to build green data centers has reduced costs by 40%, which is a common practice in the industry.
2. Is AI a bubble? Alibaba says "definitely not"
Many are concerned about the high costs, but Wu reassures them:
- There is a real demand for computing power: The industry believes that by 2030, there will be a shortage of AI computing power, similar to the current shortage of charging stations for electric vehicles. Those who invest early will have a competitive advantage.
- Investments are likely to pay off: Alibaba estimates that the costs of building AI infrastructure will be recouped within 3 years, and as the profit margins of AI products increase (e.g., through AI-powered advertising and customer service), the investment will pay off in 2-2.5 years. This is much faster than returns in many other industries, indicating that it is driven by real demand, not a bubble.
- Alibaba is transforming into an AI company: It is no longer just an e-commerce or internet company; it is becoming an AI company where computing power is essential, just like machinery in a factory, which requires significant investment.
3. How crazy is the global AI spending war?
Not only Alibaba but also other domestic and international giants are investing heavily in AI:
- Domestic players: ByteDance plans to invest $70 billion in AI data centers by 2026; Tencent's quarterly capital expenditures reached 52.8 billion yuan, a record for Chinese internet companies.
- International players: Microsoft, Amazon, Google, Meta, and others plan to invest over $725 billion (about 5 trillion yuan) in AI infrastructure by 2026; Google has even issued Australian dollar bonds to fund these efforts.
- Upstream players: NVIDIA, in collaboration with Wall Street giants like Goldman Sachs and Blackstone, has launched a $500 billion AI infrastructure financing initiative, recognizing that those who build the most computing power will control the pricing power in the AI era.
4. The AI race ultimately boils down to "energy"
AI is energy-intensive; a large AI data center consumes as much electricity as a small town. While money can buy chips, it cannot buy unlimited power, making energy a new competitive factor:
- Green energy is crucial: Tencent and Envision Energy's green data centers in Inner Mongolia use wind and solar power, reducing costs by 40%. Envision is also building the world's largest zero-carbon AI park in Ulanqab, with a capacity of over 2GW (enough for 2 million households).
- The competition focuses on comprehensive energy capabilities: In addition to power sources, factors such as energy costs, grid stability, energy storage (for power outages), and cooling (since AI devices generate a lot of heat) are critical for the sustainability and profitability of data centers.
5. What if Chinese companies do not join in?
American companies have already gone all in on AI. If Chinese companies do not follow suit:
- They may lose their leading position: AI infrastructure (computing power) is like a highway; if the US builds it first, Chinese companies will be forced to follow their routes or face limitations.
- They may miss out on future industry opportunities: AI will penetrate all industries, and without access to computing power, Chinese companies will miss out on market shares.
In short, this is not a matter of choice but a necessary "arms race." The competition will ultimately depend on a balance of capital, computing power, and energy. Those who can manage these factors well will emerge as winners in the AI era.
In one sentence
AI has evolved from a technology competition to an infrastructure competition, with global giants investing heavily in computing power and energy. Alibaba's $80 billion in financing is just a microcosm of this broader battle. In the coming years, those who can master computing power and energy will become the "infrastructure kings" of the AI era.