NVIDIA

NVIDIA Corporation (NASDAQ: NVDA) is the dominant designer of graphics processing units (GPUs) and AI compute infrastructure. Founded in 1993 by jensen-huang, Chris Malachowsky, and Curtis Priem; headquartered in Santa Clara, California. As of early 2026, NVIDIA is one of the most valuable companies in history, with a market capitalisation of approximately $4 trillion.

Business Model

NVIDIA is a fabless semiconductor company: it designs chips but outsources manufacturing to tsmc. Revenue sources:

  • Data Centercuda-enabled GPUs for AI training and inference (dominant and growing segment)
  • Gaming — GeForce consumer GPUs; still NVIDIA’s primary consumer brand
  • Professional Visualisation — Quadro/RTX workstation GPUs
  • Automotive — DRIVE platform for autonomous vehicles
  • OEM and other — embedded and edge devices

The AI Factory Era

NVIDIA now positions itself as an AI computing platform company rather than a chip company. Its products span the full stack: cuda (software platform), GPUs, CPUs (Grace/Vera), networking (NVLink, InfiniBand via Mellanox acquisition), and complete rack-scale systems (DGX, HGX, Vera Rubin pods). The unit of sale has expanded from chip → server → rack → ai-factory. See rack-scale-computing.

CUDA Moat

cuda is NVIDIA’s primary competitive advantage — not the transistors, but the developer install base, library ecosystem, and 20-year trust relationship with researchers and engineers globally. See cuda.

Key Products Referenced

  • GeForce — consumer gaming GPU; carrier of CUDA into homes and universities
  • DGX / HGX — data-centre AI compute systems
  • NVLink 72 — rack-scale GPU interconnect enabling trillion-parameter model inference
  • Vera Rubin pod — next-generation agentic-ai rack system (2026)
  • CUDA — parallel computing platform and API, version 13.2 as of 2026
  • Nemotron 3 Super — 120B-parameter open-weight MoE model; open data + recipe
  • DLSS 5 — AI-powered game rendering upscaler
  • RTX Mod — AI modding tool for classic games

Supply Chain

~200 suppliers contribute to the 1.3–1.5 million-component Vera Rubin rack. Key upstream partners: tsmc (fabrication), SK Hynix (HBM memory), ASML (EUV lithography tooling). NVIDIA manages supply chain through multi-year demand signalling and personal CEO-to-CEO relationship management.


Sources: fridman-huang-2026-nvidia-ai-revolution