Summary of Key Points
NVIDIA has recently launched two major strategic moves: first, in collaboration with six top Wall Street financial institutions (such as Blackstone and Bridgewater), it has established a $500 billion special fund to assist customers in financing the purchase of chips and building data centers; second, it invested $3 billion in the power company Lancium to secure gigawatt-level electricity resources necessary for AI data centers. The underlying logic is that the competition in AI computing power has shifted from "grabbing chips" to "grabbing electricity"—without stable power supply, even the most advanced chips are useless. NVIDIA is transitioning from simply selling hardware to controlling the "lifeblood" of the AI ecosystem: finance (ensuring customers can afford chips) and energy (ensuring chips can operate efficiently).
Detailed Analysis
1. The $500 Billion Fund: Turning Chips into Profit-Generating Assets
The fund established by NVIDIA in partnership with these financial giants essentially provides loans for customers to purchase chips. Previously, companies had to pay a large sum of money upfront for AI chips (for example, hundreds of millions for GPUs). Now, they can use future computing power revenues as collateral to borrow from the fund to buy chips and build data centers. NVIDIA positions GPUs as assets that generate continuous cash flows (e.g., customers using the computing power to earn revenue to repay loans), making it more attractive for financial institutions to provide financing.
For NVIDIA, this approach has several benefits: customers no longer face barriers due to lack of funds, leading to a surge in sales; moreover, NVIDIA can profit from the financial transactions (such as interest income), rather than relying solely on one-time hardware sales. The Wall Street giants are also supportive because computing power is becoming a core asset akin to oil, offering stable returns.
2. Investing $3 Billion in Electricity: Power Supply as the Next Scarce Resource
Chips require substantial amounts of electricity to operate—the power consumption of an AI data center can be equivalent to that of a small town. NVIDIA's investment in Lancium is like securing access to "fuel stations" for these data centers. Lancium owns approved power resources (e.g., 4 gigawatts in Texas, sufficient to power hundreds of thousands of GPUs) and is also involved in building power infrastructure for companies like OpenAI and Microsoft.
Why is electricity so crucial? Jensen Huang has stated, "Chips without power are just inventory; they’re not intelligent." Many regions in the U.S. are restricting data center power usage (for example, the Texas governor is reviewing data center applications due to their high demand for new power). Companies that can secure stable power supply gain a significant advantage in the AI industry. By investing in Lancium, NVIDIA provides customers with a package of chips and electricity, ensuring they can use them without worrying about power shortages.
3. Jensen Huang's "Five-Layer Cake" Model: Energy as the Foundation of AI
Huang’s five-layer model for AI places energy at the foundation:
- Layer 1 (Foundation): Energy (electricity, renewable sources);
- Layer 2: Chips (GPUs, etc.);
- Layer 3: Infrastructure (data centers);
- Layer 4: AI models (ChatGPT, etc.);
- Layer 5: Industry applications (finance, healthcare, etc.).
Without energy, none of the other layers can be established. Training a large AI model consumes millions of kilowatt-hours of electricity—equivalent to what a household uses over several decades. AI is not an abstract concept; it’s a physical process where electrons are processed in chips, making electricity indispensable.
4. The U.S. Power Crisis: AI's Impact on Resource Competition
The demand for data center power in the U.S. is skyrocketing—90% of new power applications from ERCOT (Texas Electric Reliability Corporation) come from data centers, representing five times the state’s historical peak usage. As a result, many regions are restricting new data center construction (the number of restrictions increased from 300 to over 500 in July). Companies with approved power resources become highly valuable. NVIDIA's investment in Lancium gives it access to these resources, ensuring its customers can use chips without power concerns.
5. NVIDIA's Ambition: From Hardware Supplier to Controller of the AI Ecosystem
NVIDIA used to be a manufacturer of GPUs but now aims to become the coordinator of the entire AI ecosystem:
- Financially: It finances customers and binds them to its products (they can only buy NVIDIA chips);
- Energetically: It ensures customers rely on its power solutions;
- Technologically: Its GPUs are essential for AI, so customers are already dependent on it.
By doing so, NVIDIA earns profits from all aspects of the AI industry: chip sales, finance, and energy. Its goal is to make every AI player dependent on it—without its chips, power, or financing, they cannot operate effectively.
Conclusion
The next battle in AI is no longer about who has the most chips but who can supply the necessary power. NVIDIA is using financial and energy strategies to build a new barrier of protection, enabling more people to access chips while ensuring their efficient operation. This move transforms the competition from a technical one into an infrastructure one, with NVIDIA taking the lead.