Nvidia Unveils New AI Hardware to Expand Global Reach and Flexibility

Nvidia CEO Jensen Huang launched a range of new technologies at Computex in Taiwan to support the continued global growth of artificial intelligence. Speaking at Asia’s leading electronics event, Huang unveiled upgrades to Nvidia’s AI ecosystem, including the upcoming GB300 systems designed to handle complex AI workloads. These next-gen systems, expected in the third quarter of 2025, will improve upon the current Grace Blackwell AI models already being deployed by top cloud service providers.

A major highlight was Nvidia’s introduction of NVLink Fusion systems, which allow more flexibility for data centers. For the first time, Nvidia will enable customers to mix and match their own CPUs with Nvidia’s AI chips, or pair Nvidia CPUs with third-party AI accelerators. This shift opens up Nvidia’s hardware architecture, offering greater customization while keeping Nvidia’s technology central to performance.

The move comes as companies like Microsoft and Amazon develop their own processors, which could reduce reliance on Nvidia’s hardware. To counter this, Nvidia is collaborating with key industry players. MediaTek, Marvell, and Alchip will design custom AI chips compatible with Nvidia systems, while Qualcomm and Fujitsu will build processors to work alongside Nvidia’s accelerators.

These developments aim to strengthen Nvidia’s leadership in the AI hardware space and promote broader adoption across industries and regions.

Source: Business Standard