Summary of Key Points
NVIDIA, in collaboration with six major financial giants on Wall Street (such as JPMorgan Chase and Goldman Sachs), plans to invest $500 billion in building the infrastructure for artificial intelligence (AI). The key to this strategy lies in the “money circulation”: Wall Street provides funding for AI infrastructure companies, which in turn must purchase NVIDIA’s chips. NVIDIA reaps huge profits, allowing Wall Street to profit from both the infrastructure projects and NVIDIA’s earnings. The entire process is akin to “the people who sell shovels during a gold rush printing their own money” – NVIDIA occupies a central position, with every link in the chain relying on it, essentially earning profits without much effort.
Detailed Breakdown
1. What exactly is this $50 billion in AI infrastructure?
It’s not about building roads and bridges; rather, it’s about the “hardware foundation” for running AI:
- Super-large data centers: Places that store massive amounts of data required for AI and host large models like ChatGPT.
- High-performance chips: NVIDIA’s GPUs, which can be considered “specialized super-brains” for AI; without them, AI models cannot function.
- Supporting networks: High-speed networks that connect data centers to users to ensure smooth AI services.
Wall Street’s money will go to companies that build these facilities, but the first step for these companies is to buy NVIDIA’s chips – just as you need a pot to cook, NVIDIA’s GPUs are the “pots” for AI.
2. Why is this called a “money-printing” cycle?
Here’s an example from the 19th-century American Gold Rush: most people who dug for gold didn’t make money, but those who sold shovels and water did. In this case, NVIDIA plays the role of the shovel seller:
- Step 1: Wall Street provides funding to AI infrastructure companies (like giving start-up capital to gold miners).
- Step 2: Infrastructure companies buy NVIDIA’s chips (gold miners buy shovels).
- Step 3: NVIDIA receives the money (it has made a profit from selling shovels).
- Step 4: Infrastructure companies use the chips to provide AI services and then distribute profits to Wall Street (gold miners share their earnings with investors).
- Step 5: As NVIDIA’s profits increase, its stock price rises, allowing Wall Street to make more money on its investment.
In this cycle, NVIDIA is the most stable party – regardless of whether AI companies are profitable or not, they need to buy its chips, effectively printing money for itself.
3. Why can NVIDIA be the “shovel seller”?
Not just anyone can sell shovels; NVIDIA has two key advantages:
- Technological monopoly: Its GPUs were originally designed for gaming but turned out to be ideal for AI computing (AI processes large amounts of data quickly, and GPUs are more than 10 times faster than regular computer CPUs). Almost all major AI models (like ChatGPT and Wenxin Yiyán) use NVIDIA’s chips for training.
- Ecosystem lock-in: NVIDIA has developed a software platform called CUDA that developers use to write AI programs, which can only run on NVIDIA chips. It’s like buying an Apple phone; you can’t use other chips with the same programs, so AI companies are reluctant to switch.
4. What impact does this have on ordinary people?
- Positive aspects: As AI infrastructure develops, AI services will become cheaper and more widespread. For example, using AI for medical diagnoses (quickly analyzing CT scans), homework assistance, or content writing could become as common as using WeChat. More jobs related to AI (such as data center maintenance and AI engineering) will emerge.
- Negative aspects: NVIDIA’s dominance may lead to price increases; if chip prices rise, the cost of AI services will also increase, affecting consumers. If AI development doesn’t meet expectations, the $50 billion investment could be wasted, potentially causing financial fluctuations, though the impact on ordinary people would likely be limited.
5. Are there any risks?
- Dependency risk: All AI infrastructure relies on NVIDIA’s chips. If its production capacity falls or its technology is surpassed (for example, by Chinese companies like Huawei or American companies like AMD), the entire AI ecosystem could be disrupted.
- Bubble risk: Is $50 billion too much? If AI applications don’t become as widespread as expected, the data centers might remain idle, wasting Wall Street’s investment.
- Monopoly risk: NVIDIA holds over 80% of the AI chip market share and could face anti-monopoly investigations (such as fines or price restrictions from authorities like the US or EU).
In Conclusion
NVIDIA’s move essentially positions it as the “King of Infrastructure” in the AI era – other companies that use AI will have to pay for its services, allowing NVIDIA to profit from the entire industry. For ordinary people, this means that AI will become more accessible, but the benefits will flow primarily to NVIDIA and Wall Street.