Overview
- Nvidia reported $96.2 billion in fiscal Q2 2027 revenue with $89 billion from data-center sales in a report this week, a surge driven by demand for its high-performance GPUs.
- The company said its Grace CPU line has generated about $5 billion over the trailing 12 months and that its Vera CPUs are now in full production, signaling a push from standalone GPUs to combined CPU‑GPU platforms.
- Major customers are still buying large amounts of Nvidia hardware but are changing procurement posture, with OpenAI’s CFO saying the group is diversifying suppliers while committing roughly 12 gigawatts of Nvidia compute through 2030.
- Industry observers draw a clear split between training and inference workloads: training favors Nvidia’s general‑purpose GPUs for episodic model builds, while continuous inference can favor purpose‑built inference chips and open spots for competitors.
- OEM order books and backlogs, including Dell’s roughly $95 billion AI-server backlog tied to Nvidia platforms, show sustained customer demand even as investor, regulatory, and AI‑safety concerns introduce new business risks.