NVIDIA
The dominant force in AI computing — powering 94% of the data center GPU market with $130.5B in revenue, 75% gross margins, and the CUDA software moat that no competitor has cracked.
Business Overview
What does NVIDIA do, and why is it dominating AI?
NVIDIA designs the brains behind artificial intelligence. Every time ChatGPT answers a question, every time a self-driving car processes its surroundings, every time a drug company simulates molecular interactions — there is a very high chance that NVIDIA hardware is doing the computation. The company has evolved from a graphics card maker for gamers into the most critical infrastructure provider in the AI revolution.
Founded in 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem, NVIDIA spent decades perfecting GPU (Graphics Processing Unit) architecture. The pivotal insight was that GPUs — originally designed to render video game graphics — are perfectly suited for the parallel processing that AI and machine learning demand. That bet, placed over 20 years ago with the creation of the CUDA software platform, is now paying off spectacularly.
NVIDIA's true competitive advantage is not just hardware — it is the CUDA software ecosystem. Over 4 million developers write code optimised for NVIDIA's GPUs. Libraries like cuDNN, TensorRT, and RAPIDS form the backbone of modern AI development. Switching to a competitor means rewriting millions of lines of code, retraining engineering teams, and accepting inferior performance. This is why even companies building their own custom AI chips (Google TPUs, Amazon Trainium) still rely heavily on NVIDIA GPUs.
The latest Blackwell architecture (GB200/GB300 NVL72) is already in production, offering a massive leap in AI training and inference performance. With 40,000+ companies building on the NVIDIA platform, the company sits at the centre of the most transformative technology shift since the internet.
Data Centre
The growth engine — AI training & inference GPUs, DGX systems, networking (NVLink, Spectrum-X), and NVIDIA AI Enterprise software. Over 85% of total revenue.
Gaming
GeForce RTX GPUs for PC gaming with 80%+ discrete GPU market share. Ray tracing, DLSS, and AI-enhanced graphics keep the brand premium.
Auto, Robotics & Omniverse
DRIVE platform for autonomous vehicles, Isaac for robotics, and Omniverse for industrial digital twins. Early-stage but high-growth verticals.
Financial Fundamentals
Three tests every quality business must pass
Moat Analysis
Five dimensions that determine competitive durability
Brand Loyalty & Pricing Power
9/10NVIDIA commands 75% gross margins on hardware — unheard of in semiconductors. The brand is synonymous with AI computing. 94% data centre GPU market share and 80%+ gaming GPU share means customers willingly pay premium prices. 40,000+ companies run on NVIDIA technology.
High Barriers to Entry
9/10The CUDA ecosystem took 20+ years and $100B+ in cumulative R&D to build. Competitors like AMD (ROCm) remain 2–3x behind in software maturity. Even Intel abandoned discrete GPU competition. Building a comparable chip-to-software full-stack is nearly insurmountable for new entrants.
High Switching Costs
9/10CUDA lock-in is real. Code written for CUDA cannot run on AMD or Intel GPUs without costly re-engineering. Tech giants have invested billions in NVIDIA-based data centres. Switching means rewriting hundreds of thousands of lines of code, retraining teams, and accepting months of disruption.
Network Effect
7/104 million+ CUDA developers create a powerful flywheel: more developers write more CUDA libraries, making the platform more valuable, attracting more developers. Academic institutions train students on CUDA, reinforcing the cycle. However, it is an indirect B2B network effect rather than a direct consumer platform.
Economies of Scale
8/1094% market share drives manufacturing scale, better TSMC wafer pricing, and massive R&D amortisation across a huge install base. $130.5B revenue means NVIDIA can outspend any competitor on R&D while maintaining 75% gross margins. However, NVIDIA relies on TSMC for fabrication, limiting some scale advantages.
Bull & Bear Thesis
Both sides of the coin — so you can decide for yourself
Bull Case
Bear Case
Growth Drivers
Where the next wave of revenue comes from
Blackwell Architecture Ramp
The Blackwell GPU platform (GB200/GB300 NVL72) is already in production with demand far exceeding supply. It delivers a 4x performance leap in AI training and a 30x improvement in inference over the prior Hopper generation. Every major hyperscaler has placed orders.
AI Inference Market Expansion
While AI training drove the first wave, inference (running AI models in production) is a much larger long-term market. Every chatbot response, every AI search result, every autonomous driving decision requires inference compute. NVIDIA's TensorRT and inference-optimised GPUs are positioned to capture this.
Sovereign AI & Enterprise
Nations are building their own AI infrastructure for data sovereignty. NVIDIA is partnering with governments worldwide to build national AI supercomputers. Meanwhile, enterprise adoption is just beginning — most of NVIDIA's $130.5B revenue comes from a handful of hyperscalers. Enterprise AI is the next massive wave.
Robotics & Autonomous Vehicles
The NVIDIA DRIVE platform powers autonomous vehicle development at Toyota, Mercedes, and dozens of others. Isaac for robotics and Omniverse for digital twins are early but high-potential verticals. As physical AI matures, these segments could become multi-billion dollar revenue streams.
Investment Risks
Every investment has risks — here is what could go wrong
Customer Concentration
A handful of hyperscalers (Microsoft, Google, Amazon, Meta) account for a massive share of NVIDIA's data centre revenue. If even one major customer shifts to custom silicon or reduces capex, the revenue impact could be significant.
High SeverityGeopolitical & Export Restrictions
U.S. government export controls have already restricted NVIDIA's sales to China. Further tightening — or retaliatory measures — could cut off access to one of the world's largest AI chip markets and disrupt global supply chains.
High SeverityAI Capex Cycle Slowdown
If AI infrastructure spending enters a digestion phase — where companies pause to utilise what they have already built — NVIDIA's revenue growth could decelerate sharply. Semiconductor cycles have historically been brutal, and 114% growth is not sustainable forever.
Medium SeverityTSMC Single-Source Risk
NVIDIA depends entirely on TSMC for chip fabrication. A natural disaster in Taiwan, a geopolitical conflict, or severe TSMC capacity constraints would directly impact NVIDIA's ability to produce and deliver GPUs. This is an existential supply chain risk.
Medium SeverityValuation & Intrinsic Value
What is this business actually worth?
As of 16 February 2026, NVIDIA (NVDA) is trading at approximately 13% below its estimated intrinsic value based on our discounted cash flow model. With $130.5B in revenue, $60.85B in free cash flow, 165% ROIC, a fortress balance sheet, and dominance across the most transformative technology shift in a generation, the current price offers a modest margin of safety for long-term investors.
Download the Mintin5 app for the latest update on valuation details, fair value estimates, and margin of safety calculations for NVDA and 100+ other stocks.
Download Mintin5 for Valuation DetailsDisclaimer
This research is for educational purposes only and does not constitute financial advice. The information presented is based on publicly available data and our independent analysis. Always do your own research and consult a qualified financial advisor before making any investment decisions. Past performance is not indicative of future results.