NVIDIA Corp
Designs GPUs and networking systems that power AI infrastructure and data center computing.
What they do
NVIDIA designs graphics processing units (GPUs) originally built for gaming but now the backbone of AI training and inference. The company also manufactures networking hardware and complete data center systems used by hyperscalers and enterprises building AI infrastructure. Gaming, professional visualization, automotive, and data center represent their core markets.
How they make money
Data center revenue dominates, driven by sales of GPU accelerators (H100, H200, Blackwell series) and networking gear to cloud providers and AI labs. Gaming GPU sales to consumer and OEM channels provide a secondary revenue stream. Professional visualization and automotive chips contribute smaller but growing shares.
The numbers
Recent filings show NVIDIA filing its FY2026 10-K on February 25, 2026, covering fiscal year ending January 25, 2026. The company trades at 40.7x trailing P/E but only 23.9x forward P/E, suggesting analysts expect sharp earnings growth. At $199.88, the stock sits 5.8% below its 52-week high of $212.19, indicating recent price consolidation. The $4.9 trillion market cap reflects investor belief that AI infrastructure spending remains durable despite the stock's pullback from all-time highs.
Price action
NVDA trades at $199.88, down from a 52-week high of $212.19 but more than double its 52-week low of $95.04. The stock has consolidated near current levels after a powerful rally, with the 6% drawdown from highs suggesting profit-taking rather than fundamental deterioration. Forward P/E compression from 40x to 24x indicates the market is pricing in aggressive earnings growth over the next twelve months.
- 01AI infrastructure buildout remains in early innings, with hyperscalers and enterprises racing to deploy GPU capacity for training and inference workloads that only NVIDIA's architecture can handle at scale.
- 02CEO Jensen Huang characterizes manufacturing bottlenecks as a '2-3 year problem' that proves demand far exceeds supply, protecting pricing power and implying multi-year visibility (April 2026 comments).
- 03Forward P/E of 23.9x trades at a discount to trailing 40.7x, suggesting analysts model significant earnings expansion as Blackwell ramps and networking attach rates increase.
- 01Mega-cap valuation at $4.9 trillion leaves little room for execution missteps, with any demand slowdown or competitive threat likely to trigger sharp multiple compression.
- 02Manufacturing bottlenecks acknowledged by CEO could delay revenue recognition and allow competitors like AMD or custom hyperscaler chips to gain share during supply constraints.
- 03Forward P/E of 24x assumes flawless execution on next-gen products and sustained hyperscaler capex, both vulnerable to macro shocks or shifts in AI investment priorities.
Upcoming catalysts
- ▸Blackwell GPU family production ramp and customer deployment timelines, which determine whether NVIDIA can convert its enormous backlog into recognized revenue.
- ▸Hyperscaler capital expenditure guidance from Microsoft, Google, Meta, and Amazon, which collectively represent the majority of data center GPU demand.
- ▸Competitive threats from AMD's MI300 series, custom AI chips from Google (TPU) and Amazon (Trainium), or any signs of architectural alternatives gaining traction.
- ▸Export control changes that could restrict sales to China or other geographies, impacting addressable market and revenue mix.
Questions to ask yourself
- “How sustainable is gross margin above 70% if competition intensifies or hyperscalers negotiate harder as they build leverage through custom silicon?”
- “What percentage of revenue comes from the top five customers, and how vulnerable is NVIDIA if any single hyperscaler slows AI infrastructure spending?”
- “How does NVIDIA plan to address the 2-3 year manufacturing bottleneck without losing share to competitors who solve supply chain faster?”
Risks often missed
- ⚠Concentration risk in customer base and end markets, with data center sales to a handful of hyperscalers representing the majority of revenue growth (10-K risk factors).
- ⚠Dependence on third-party foundries, primarily TSMC, for all chip manufacturing creates supply chain vulnerability and limits control over production ramps (10-K risk factors).
- ⚠Rapid product cycles require continuous R&D investment to maintain architectural leadership, with any stumble potentially allowing competitors to close the performance gap (10-K risk factors).