Nvidia CEO Jensen Huang announced numerous announcements on Monday and revealed new products designed to keep the company at the center of AI development and computing.

One of the most famous announcements is its new “NVLINK Convergence” program, which will allow customers and partners to use non-NVIDIA central processing units and graphics processing units and graphics processing units with NVIDIA’s products and their NVLink.

So far, NVLINK has closed the chips made by NVIDIA. NVLink is a technology developed by NVIDIA to connect and exchange data between its GPU and CPU.

“NV Link Fusion is meant to enable you to build semi-regular AI infrastructure, not just semi-regular chips,” Huang said at Computex 2025 at Asia’s largest electronic conference.

According to Huang, NVLINK convergence allows AI infrastructure to be combined with different CPUs and application-specific integrated circuits (ASICs). “Anyway, you have the benefits of using NV link infrastructure and the NV Link ecosystem.”

NVIDIA announced on Monday that NVLink Fusion's AI chip production partners have included Mediatek. Malville,,,,, alchemy,,,,, Astera Labs,,,,, concept and Rhythm. NVIDIA customers like under NVLink convergence Fujitsu and Qualcomm Technology It added that it will also be able to connect its own third-party CPUs to NVIDIA's GPUs in AI data centers.

According to Ray Wang, a semiconductor and technology analyst in Washington, NVLink represents NVIDIA's plan to capture ASIC-based data centers that have traditionally been seen as NVIDIA competitors.

Although NVIDIA dominates GPUs for general-purpose AI training, many competitors see room for chip expansion designed for more specific applications. Some of NVIDIA's biggest competitors in AI computing (and its largest customers) include cloud providers such as Google, Microsoft and Amazon, all of which are building their own custom processors.

Wang said NVLink Fusion "brings NVIDIA as the center of the next-generation AI factory, even if these systems are not fully built with Nvidia chips," Wang said, noting that NVIDIA has opened up customers for NVIDIA that do not support full NVIDIA-based NVIDIA systems, but is seeking certain GPUs.

"If adopted widely, NVLINK convergence could expand NVIDIA's industry footprint by working deeper with custom CPU developers and ASIC designers to build future AI infrastructure," Wang said.

However, according to Rolf Bult, an equity research analyst at New Street Research, NVLink Fusion does have the potential to reduce demand for NVIDIA CPUs by allowing NVIDIA customers to use alternatives.

However, “at the system level, the added flexibility increases NVIDIA’s GPU-based solutions competitiveness with alternative emerging architectures, helping Nvidia maintain its position at the center of AI computing.”

NVLINK converged ecosystem does not exist in NVIDIA's competitors Broadcom, AMD and Intel.

Huang delivered his keynote and updated his AI workload to NVIDIA's next-generation Grace Blackwell Systems. He said the company's "GB300" will be released in the third quarter of this year and will provide higher overall system performance.

On Monday, NVIDIA also announced the new NVIDIA DGX Cloud Lepton, an AI platform with a computing market, which NVIDIA said would connect world AI developers to a global network from cloud providers.

"DGX Cloud Lepton ensures a critical challenge for reliable, high-performance GPU resources by unifying access across the NVIDIA Compecute ecosystem," the company said in a press release.

In his speech, Huang also announced plans for a new office in Taiwan, where it will also establish an AI supercomputer project with Taiwan's Foxconn, which is officially known as the world's largest electronics manufacturer.

Huang said: “We are excited to work with Foxconn and Taiwan to help build Taiwan’s AI infrastructure and support TSMC and other leading companies to drive innovation in the AI ​​and robotics era.”