Summary of Key Points
NVIDIA's financial report for the second quarter of fiscal year 2027 exceeded expectations (revenue of $96.2 billion, a year-on-year increase of 106%; data center revenue of $89 billion, a year-on-year increase of 117%). The guidance for the next quarter of $10.8 billion also surpassed market expectations, with revenue expected to grow by 70% in the middle term (fiscal year 2028). However, the stock price initially fell before rising after the report was released. The reason is that "exceeding expectations" on a single quarter is no longer novel; investors are more concerned with the long-term growth story. NVIDIA no longer just sells GPU hardware; it is deeply involved in the AI industry chain through various methods such as "hardware + cloud revenue sharing," providing financial support to customers, and securing upstream supply chains in advance, in an attempt to address customers' financing and capacity bottlenecks. This also exposes it to new risks (customer credit, supply chain commitments). Ultimately, the market recognized its long-term growth prospects, and the stock price rebounded.
Detailed Analysis
1. The Financial Report Was Impressive, but the Stock Price First Fell Then Rose: Exceeding Expectations Has Become the Norm; Investors Want a "Longer-Term Story"
NVIDIA's financial results this time were almost perfect: both revenue and data center revenue doubled, and the guidance for the next quarter exceeded market expectations. So why did the stock price fall after the report was released? Over the past eight quarters, NVIDIA has consistently exceeded expectations, so a single quarter's better performance is no longer enough to surprise investors. They are now interested in how much more it can grow in the future. It was only when management mentioned the ambitious growth target of 70% for fiscal year 2028 during the conference call that the stock price turned upward (more than 4% after the market closed and another 8% before the next day's opening). In short, for NVIDIA, relying on a good quarter's performance alone is not enough; it must prove that it can continue to perform even better in the coming years.
2. From a One-Time Transaction to Long-Term Revenue Generation: GPU Sales Are No Longer a One-Time Profit
In the past, NVIDIA's GPU sales were a one-time deal. Now, it aims to generate continuous revenue from these GPUs:
- New Model: After selling GPUs to AI cloud service providers, NVIDIA also gets a share of the cloud revenue generated from using those GPUs (for example, if a customer uses the GPUs to provide computing services and makes money, NVIDIA gets a portion of that revenue).
- Example: The 5-year, $3.4 billion contract signed with IREN in 2026 is a model that combines hardware sales with rental revenue sharing.
- Significance: This transforms a one-time transaction into a long-term partnership, stabilizing future revenue streams and making it more attractive for customers, as NVIDIA becomes financially aligned with them.
3. From a Chip Manufacturer to a "Half-Bank": Helping Customers Solve Financial Issues to Secure More Orders
AI customers (such as small AI cloud companies and model development firms) may want to buy GPUs but lack the funds to build data centers or pay for electricity. NVIDIA steps in to help:
- Credit Support: It provided $105 billion in credit support for OpenAI's data center in Ohio (not a direct payment, but a guarantee that the customer can obtain a loan, to be effective in phases).
- Financing Platform: NVIDIA collaborates with firms like Blackstone and Goldman Sachs to create a computing financing platform, aiming to attract $500 billion in third-party capital to help customers purchase GPUs.
- Data Center Rental: In July, NVIDIA signed a 15-year contract to rent a 1-gigawatt data center in Texas, which can accommodate hundreds of thousands of GPUs, and then subleases it to partners, solving the problem of customers not having their own data centers.
Jensen Huang described this as not just "internal capital circulation" but an investment necessary for the transformation of the AI platform. Essentially, it uses its own credit and resources to enable customers to purchase GPUs, which in turn converts into orders for NVIDIA.
4. Securing the Upstream Supply Chain: Locking in Capacity, But Also Taking on Risks
With strong AI demand and insufficient GPU capacity, NVIDIA takes proactive steps to secure supply:
- Data: The supply commitment for this quarter increased from $119 billion to $279 billion (an increase of $160 billion in three months, equivalent to 1.66 times the quarterly revenue), mainly due to increased memory purchases.
- Examples: NVIDIA partnered with Amkor, a packaging and testing company, to support its capacity expansion in the United States; it also has long-term agreements with TSMC and SK Hynix to ensure the supply of wafers and memory.
- Risks: If AI demand suddenly declines, these pre-locked supply chain commitments could lead to inventory buildup or losses, as NVIDIA has essentially "written its assumptions about AI demand into the contracts."
5. The Double-Edged Sword of the New Model: A Deeper Moat, but Also Greater Risks
NVIDIA's new model makes it harder for competitors to replace it, but it also brings new pressures:
- Benefits: A deeper moat, making it harder for competitors to challenge NVIDIA. They would need to not only produce chips with similar performance but also have the capability to secure supply chains, provide financing, and build ecosystems, which is not easy for smaller companies to achieve.
- Risks:
1. Customer Credit Risk: If customers (such as AI companies) fail to make money and cannot repay the loans, NVIDIA's guarantees could be at risk.
2. Upstream Commitment Risk: If demand decreases, the pre-locked supply chain commitments could become a burden.
3. Customer Diversification: AI companies like Anthropic are starting to buy AMD chips, so NVIDIA's customer base is not solely dependent on it. Its financial support does not guarantee permanent exclusivity.
Jensen Huang emphasized that while customers have a real demand for computing power, "demand does not equate to profitability." NVIDIA must prove that customers who buy its GPUs can continue to make money and repay their loans.
In Conclusion
NVIDIA has evolved from a chip manufacturer to an organizer of the AI industry chain. It has used its resources to address the bottlenecks in AI development, but it has also tied its own fate more closely to that of the entire AI industry. The future potential of NVIDIA depends on whether its customers can truly generate revenue.
(End of the translation.)