NVIDIA Q3 earnings report fuels AI frenzy: Bubble theory vanishes, how can the leader remain strong forever?

After the U.S. market closed on the 19th, $NVIDIA(NVDA.US) released its Q3 2026 fiscal year earnings report.

Revenue reached $57.006 billion, up 62% year-over-year; net profit was $31.91 billion, up 65.3% YoY; EPS was $1.30, up 60%. The even brighter Q4 guidance—$65 billion in revenue.

CEO Jensen Huang stated bluntly: "All cloud GPUs are sold out, and demand for Blackwell far exceeds expectations."

This earnings report not only "slapped" the rampant AI bubble narrative but also ignited investors' lasting enthusiasm for AI leaders. The AI wave is far from cooling down.

NVIDIA's Q3 earnings report can be called the semiconductor industry's "Manhattan Project." The core driver is the data center business, with revenue reaching $51.2 billion, up 66% YoY, accounting for nearly 90% of total revenue. This isn't just a pile of numbers but an exponential explosion in AI training and inference demand: from $2.9 billion four years ago to $51.2 billion today, a 17-fold increase.

In contrast, gaming and AI PC businesses, though stable at $4.26 billion with 30% YoY growth, have become "supporting roles"; professional visualization (Pro Viz) and automotive (Auto) segments contributed $760 million and $592 million, up 56% and 32%, respectively, showing AI's penetration into edge computing.

The highlight lies in gross margin resilience: despite H20 chip exports to China contributing only $50 million due to geopolitical restrictions, non-GAAP gross margin remained at 73.6%, down just 0.4 percentage points. R&D spending of $4.71 billion, up 39%, demonstrates NVIDIA's sustainable "burn rate" on AI infrastructure. Huang emphasized on the earnings call: "The AI ecosystem has entered a virtuous cycle, with training and inference demand accelerating exponentially." This isn't empty talk—the company revealed that the next-gen chip is expected to drive $500 billion in revenue, with the data center TAM already at the trillion-dollar level.

Market reaction was enthusiastic. NVIDIA's after-hours stock price surged over 6%, driving a collective rally in data center concept stocks, reflecting ecosystem synergy: as NVIDIA's largest GPU buyer, $Coreweave(CRWV.US) soared 9%, $Nebius(NBIS.US) rose 9%; downstream applications like Applied Digital surged 18%. This "concept stock surge" confirms the deep embedding of the AI supply chain: from chips to cloud services to edge applications, NVIDIA has become the "invisible hub."

Recently, the AI bubble narrative loomed over the market: from Microsoft and Amazon's massive CapEx to OpenAI's valuation bubble, skepticism arose—"Is AI investment a self-contained loop?"

NVIDIA's earnings report was a knockout punch, shattering this narrative. Huang's "cloud GPUs sold out" declaration isn't marketing hype but ironclad proof of supply-demand imbalance: Blackwell platform demand "far exceeds expectations," with GB300 revenue already surpassing GB200.

Deeper analysis reveals AI's "double helix" momentum: on one hand, hyperscalers like Google and Microsoft are racing on CapEx, driving training demand; on the other, inference workloads like ChatGPT's real-time responses are shifting from "nice-to-have" to "must-have."

The bubble narrative's collapse isn't luck but ecosystem moat: NVIDIA's CUDA stack has become the "English" of AI development—switching costs are astronomical.

Is the AI leader worth holding? Three Q&As

1. Valuation bubble?

Current PE (TTM) is ~52.69-55.04x, seemingly high, but forward PE drops to 27-30x based on Q4 guidance. Compared to 2021's chip cycle highs, AI's "perpetual growth" narrative is more compelling—TAM scaling from trillions to tens of trillions is infrastructure revolution, not cyclical fluctuation.

2. Competition risks?

AMD and Intel are eyeing, but NVIDIA's market share exceeds 80%. Huang's "exponential acceleration" hints Blackwell will widen the gap. China market constraints are buffered by global demand.

3. Macro uncertainty?

Ahead of the Fed's December meeting, jobs data is fuzzy, but AI as a "productivity tool" is cycle-resistant. Historical data shows NVIDIA outperformed in 2008 and 2020 downturns.

In short, the AI leader is worth holding—not blindly, but based on ironclad demand logic. "AI trade remains the top narrative, NVIDIA is its heart."

Under the trillion-dollar TAM's call, NVIDIA isn't just a stock but a mirror of AI's future. Holding the leader may not be gambling but betting on the productivity revolution.