Editorials

On August 26, NVIDIA filed its Q3 fiscal 2026 10-Q for the quarter ending October 26, 2025. The headline numbers confirm what has become unmistakable: in artificial intelligence infrastructure, NVIDIA owns the commanding position—and its pricing power is accelerating.

The quarter saw revenue of $57.0 billion, producing gross profit of $41.8 billion. That gross margin of 73.4% is the kind of profitability that distinguishes a monopoly-grade business from merely a successful one. For context: this is not a flash-in-the-pan reading. The gross margin has remained >72% across three consecutive quarters, confirming that the ultra-high-margin dynamic is structural, not cyclical.

Gross margin expansion of this magnitude, sustained this long, tells a story that cannot be rewritten by supply-chain normalization or competitive relief. In Q1 fiscal 2026 (ended April 27), NVIDIA's gross margin was 60.5%. By Q3, it had expanded 728 basis points to 73.4%—a one-year swing of extraordinary breadth.

Why? Not because costs fell. Cost of revenue as a percentage of sales compressed dramatically because NVIDIA's customers—cloud providers, hyperscalers, and enterprise AI groups—have no practical alternative. They are locked into the CUDA ecosystem, the software moat that transforms hardware into inseparable infrastructure. When your data center fleet depends on GPUs that run CUDA, you buy the next-generation NVIDIA part at the asking price, when it becomes available.

This is leverage. This is pricing power. This is the signature of a dominant franchise.

For fiscal year 2026, NVIDIA reported segment revenue as follows:

• Compute and Networking: $47.4 billion (78% of total revenue) with operating income of $32.0 billion, yielding an operating margin of 67.5%.
• Graphics: $13.5 billion (22% of total revenue) with operating income of $5.8 billion, yielding an operating margin of 43.3%.

The Compute and Networking segment—the data center business—is where NVIDIA's structural advantage lies. At $47.4 billion in annual revenue with a 67.5% operating margin, this single segment would rank as a Fortune 500 company unto itself, and one of exceptional profitability. The segment's growth is driven by AI model training and inference workloads at cloud providers (Microsoft Azure, Amazon AWS, Google Cloud), hyperscalers (Meta, OpenAI infrastructure partners), and enterprise deployments.

The Graphics segment, serving gaming and professional visualization, contributes a respectable base but operates at margins reflective of more traditional competitive markets. The data center segment, by contrast, operates in a world with one credible supplier of the critical component.

In Q2 (ended July 27, 2025), NVIDIA generated $46.7 billion in revenue. By Q3 (ended October 26, 2025), that number had grown to $57.0 billion—a sequential increase of $10.3 billion, or +22%. Gross profit grew 23.6% sequentially from $33.9 billion to $41.8 billion.

Earnings per share (basic) expanded from $1.08 to $1.31 quarter-over-quarter, an increase of 21.3%. This is not the profile of a company watching demand plateau. This is a company running to keep up with its backlog.

For Q4 (the company's fiscal year-end, January 25, 2026), revenue topped $68.1 billion with gross profit of $51.1 billion, yielding a gross margin of 75.0%—the highest of the year. Net income for the year totaled $120.1 billion on revenues of $215.9 billion, translating to a net margin of 55.6%. These are earnings metrics that would have been considered impossible just three years prior.

The sequential acceleration, quarter after quarter, into calendar Q4 2025 and fiscal Q4 2026 suggests that the AI infrastructure buildout is not a cyclical wave cresting; it is a structural shift that NVIDIA is positioned to monetize for years.

What underpins these numbers is not transient demand, but structural lock-in. NVIDIA's CUDA platform has become the de facto standard for AI workloads. Retraining models, rewriting libraries, and migrating to competing accelerators (AMD's ROCM, Intel's data center GPUs, custom Silicon like Google's TPUs) incurs switching costs measured in engineering months and operational risk measured in production delays.

This lock-in effects a pricing dynamic where the marginal customer is not choosing NVIDIA over a competitor at a lower price. The marginal customer is choosing to deploy NVIDIA at a given price or delay deployment. In a competitive market, the marginal buyer sets price. In a market with structural switching costs and no credible substitute, the supplier sets price, and the customer's choice is to buy or wait.

NVIDIA's customers are not waiting. They are buying at whatever price NVIDIA's supply and manufacturing constraints will bear. The 73% gross margin is evidence of this asymmetry in real time.

Investors watching NVIDIA should monitor three data points:

1. Gross Margin Sustainability. If the 73%+ gross margin persists through fiscal 2027 and beyond, the company has successfully crossed from cyclical windfall into structural dominance. A contraction below 70% would signal either competitive relief or demand normalization. A margin below 65% would suggest a material erosion of pricing power.

2. Data Center Segment Growth. The Compute and Networking segment is NVIDIA's core. As long as enterprise and hyperscaler AI deployments accelerate, revenue growth in this segment will remain robust. Flattening here would be a yellow light. Contraction would be a red one.

3. CUDA Ecosystem Lock-In. Watch for meaningful migration of enterprise AI workloads to competing platforms (OpenAI's use of Microsoft Cobalt chips, Google's use of TPUs, Meta's custom silicon projects). Small pilots are normal. Large-scale migrations would indicate that the switching-cost moat is weakening.

As of August 2026, none of these indicators have deteriorated. The data from Q3 and Q4 fiscal 2026 show that NVIDIA's structural advantages remain intact, its pricing power is expanding, and the AI infrastructure buildout is accelerating. This is what dominance looks like in technology.