To The Moon

Nvidia's $500 Billion Bombshell: Unpacking the New Math That Sparked the Rally

$NVIDIA(NVDA)$   CEO Jensen Huang's keynote at the recent GTC DC conference stunned the market, driving a surge in the company's stock. The most critical slide, and the direct catalyst, was the guidance for Nvidia's AI chip demand.

The Most Critical Slide

The most critical slide, and the direct catalyst, was the guidance for Nvidia's AI chip demand. The first bar on this keynote's chart showed cumulative shipments of 4 million GPUs from the previous-generation Hopper architecture between 2023-2025, contributing $100 billion in revenue. For context, Nvidia's H100 (Hopper) began shipping in October 2022, with the DGX H100 shipping in May 2023.

However, the most important bar showed a forecast of 20 million cumulative GPUs from the current Blackwell and upcoming Rubin architectures between 2025-2026. It is well-known that Blackwell and Rubin chips are dual-GPU, meaning this translates to 10 million Hopper-equivalent GPUs—a 150% increase over the prior generation. To date, 6 million GPUs (3 million Hopper-equivalent) have already shipped.

A 150% shipment increase might not sound earth-shattering. But then came the bombshell: Nvidia projects $500 billion in revenue from its Blackwell and Rubin AI chips in 2025-2026. This figure is for AI chips alone, excluding its high-growth networking business, traditional gaming segment, and the China market.

What Does $500 Billion Imply?

Nvidia began initial Blackwell deliveries in Q4 2024, delayed from Q2 due to yield issues. The first "true" volume shipments began in Q1 2025, when HPE announced it had shipped its first GB200 NVL72. Nvidia's data center revenue is split into 'Compute' and 'Networking.' Our analysis focuses on the Compute segment, which includes Hopper and Blackwell AI chips.

Conservative Scenario:

~We start calculations from Q3 FY25, when Nvidia first confirmed Blackwell revenue on its earnings call. Per financial reports, combined Compute revenue from Q3 FY25 to Q2 FY26 totaled $128.2 billion (this still includes Hopper products, making this estimate conservative).

~Based on current market consensus estimates for Q3 FY26 ($40.9B) and Q4 FY26 ($46.6B), the potential revenue space remaining for FY27 (calendar year 2026) is $284.3 billion. This figure is 31% higher than the current consensus estimate of $216.4 billion for FY27. At a 50% net profit margin, this implies $34 billion in additional profit, a sum larger than Nvidia's net profit from the previous quarter.

Optimistic Scenario:

~We start from Q1 FY26 (calendar Q1 2025), as indicated in the keynote. Per reports, Compute revenue from Q1 FY26 to Q2 FY26 was $68 billion (also including Hopper). Factoring in consensus for Q3 FY26 ($40.9B) and Q4 FY26 ($46.6B), the remaining revenue potential for FY27 is $344.5 billion.

~This is 59% higher than the $216.4 billion consensus for FY27. At a 50% net margin, this implies an additional $64.1 billion in profit, more than Nvidia's combined net profit from the last two quarters.

Summary and Risk Warning

In summary, Nvidia's first-ever GTC DC event showcased an expanding partner ecosystem and powerfully strong growth forecasts. The calculations above are based solely on Nvidia's keynote projections. Actual performance will still depend on the production capacity of $Taiwan Semiconductor Manufacturing(TSM)$   , the primary "shovel-seller" for the compute revolution.

@TigerStars  @CaptainTiger  @TigerWire  @Daily_Discussion   @Tiger_chat  @Tiger_comments  @MillionaireTiger