Overview
- Nvidia still controls an estimated 80%–90% of the AI‑accelerator market, aided by its CUDA software stack and deep startup ties.
- Latest quarterly update showed revenue of about $57 billion, up 62.5% year over year, and CEO Jensen Huang said cloud GPUs are sold out.
- Hyperscalers and chip rivals are expanding alternatives, including AMD’s 6‑gigawatt pact with OpenAI, progress in AMD’s ROCm software, Google’s TPUs, and Amazon’s Trainium, with reports of Alphabet discussing TPU sales to Meta.
- Recent bullish research cites visibility to more than $500 billion tied to Blackwell and Rubin architectures and projects strong free‑cash‑flow growth, framing Nvidia as a multiyear AI platform winner.
- Skeptical views highlight rich valuation, dependency on hyperscaler spending, and tighter U.S. export controls on China, while AMD signals readiness to pay a 15% export tax to sell a downgraded MI308 in China as early as 2026.