NVIDIA Q2 revenue hits $96.22 billion, up 106%

NVIDIA and Marvell have both benefited from the rapid expansion of AI data centers, but they compete from different positions. NVIDIA leads in AI accelerators and networking, while Marvell has built strong positions in custom silicon, optical connectivity and data center switching. Their relationship became more interesting following the 2026 NVLink Fusion partnership.

The latest quarter shows how far NVIDIA's financial muscle extends. Revenue rose 106% year over year to $96.22 billion, non-GAAP EPS jumped 120% to $2.22, and non-GAAP gross margin expanded 250 basis points to 75%. The data center business alone generated $89.02 billion — more than 92% of total sales — climbing 117% year over year and 18% sequentially on rapid deployment of Blackwell and Vera Rubin systems and growing networking demand. Beyond hardware, NVIDIA's CUDA software ecosystem creates high switching costs that make it difficult for competitors to take market share.

The comparison now hinges on valuation and risk profile rather than growth alone. NVIDIA's position spans training, large-scale inference and enterprise deployment, and four top hyperscalers — Amazon, Alphabet, Microsoft and Meta Platforms — are expected to spend roughly $720 billion to $745 billion on capital expenditures in 2026, with a large portion directed toward AI infrastructure. NVIDIA expects fiscal 2028 revenues to grow approximately 70%, though supply remains a constraint — the key variable for whether that outlook holds.