NVIDIA Corporation (NVDA)

Every major AI model in the world — ChatGPT, Gemini, Claude, Llama — was trained on NVIDIA chips. In the span of three years, NVIDIA went from a gaming graphics company to the most important semiconductor firm in the AI era.

Jensen Huang co-founded NVIDIA in 1993 at a Denny's restaurant in San Jose with two fellow engineers. Their original vision: build chips that render 3D graphics for video games faster than anyone else. The GPU (graphics processing unit) they invented was designed to perform thousands of simple calculations simultaneously — perfect for rendering pixels in parallel.

The breakthrough insight came decades later: the same parallel computing architecture that renders video game graphics also happens to be perfect for training neural networks. When deep learning exploded around 2012, researchers discovered NVIDIA's GPUs could train AI models 10-50x faster than traditional CPUs. NVIDIA leaned into this accident of history, building specialized AI hardware (A100, H100, B200 chips) and the CUDA software platform that researchers were already using.

Today NVIDIA's Data Center segment (AI chips for cloud providers and enterprises) generates the vast majority of revenue and is growing over 100% year-over-year. Their customers are the biggest companies in the world — Microsoft, Google, Amazon, Meta — spending tens of billions collectively on NVIDIA hardware to build AI infrastructure.

NVIDIA's moat is the software ecosystem (CUDA). Over 4 million developers have built their AI workflows on CUDA over 15 years. Switching to a competing chip means rewriting code, retraining models, and re-educating teams. Hardware competitors exist (AMD, Intel, Google TPUs, custom chips), but the software lock-in keeps NVIDIA dominant.

The key risk is customer concentration and cyclicality. A few hyperscale cloud companies represent the majority of demand. If AI infrastructure spending slows, or if these customers successfully build their own chips (Google's TPU, Amazon's Trainium), NVIDIA's growth rate could drop sharply. The stock prices in perfection — any stumble gets punished.