Nvidia’s record data center revenue signals rising chip demand

Nvidia reported record data center revenue of US$62.3 billion for the fourth quarter of fiscal 2026, ended January 25, 2026, up 75% year over year. Demand for AI training and inference is pushing operators toward integrated systems combining GPUs, high-bandwidth memory, networking silicon and data processing units. Buyers increasingly prioritize memory bandwidth, interconnect speed and performance per watt alongside computing throughput.

AMD is expanding its Instinct accelerator portfolio, while Intel supplies server CPUs and infrastructure processors. Broadcom and Marvell support custom AI silicon and high-speed connectivity. Amazon Web Services and Google are also advancing proprietary processors, giving cloud operators more options while pressuring external suppliers to improve cost efficiency, software compatibility and product availability. High-bandwidth memory, chiplet architectures and advanced packaging are becoming increasingly important as power and cooling constraints shape infrastructure decisions.

DataM Intelligence estimates the global data center chip market reached US$16.51 billion in 2025 and projects US$50.91 billion by 2035, with a CAGR of 11.92% during 2026-2035. North America accounted for 38.54% share in 2025, while GPUs represented 46% of the market in 2025. Asia-Pacific is identified as the fastest-growing region.