Jensen Huang, co -founder and executive director of Nvidia Corp., speaks at the door of the Computex conference in Taipei, Taiwan, on Monday, May 19, 2025.
Bloomberg | Bloomberg | Getty images
Nvidia The Jensen Huang CEO made a series of ads and revealed new products on Monday that aim to keep the company in the center of artificial intelligence development and computer science.
One of the most notable ads was its new “Nvlink Fusion” program, which will allow customers and partners to use non -NVIDIA central processing processing units and graphics processing units along with NVIDIA products and their NVLINK.
Until now, Nvlink was closed to 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 so that I can build the semi-collapsed infrastructure, not only semi-rustominous chips,” Huang said in Computex 2025 in Taiwan, the largest electronic conference in Asia.
According to Huang, NVLink Fusion allows Nvidia processors to combine NVIDIA processors and specific integrated circuits of the application (ASICS). “In any case, it has the benefit of using the NV link infrastructure and the NV link ecosystem.”
Nvidia announced Monday that AI chipping partners for nvlink fusion already includes Mediaek, Marvell” Alchip” Astera Labs” Synopsis and Cadence. Under the nvlink fusion, Nvidia customers like Fujitsu and Qualcomm technologies You can also connect your own third -party CPUs with NVIDIA GPUs in AI data centers, he added.
Ray Wang, a semiconductor analyst and technology based in Washington, told CNBC that the NVLink repeats Nvidia’s plans to capture a part of the data centers as a function of the ASIC, which have the traditional leg as competitors of Nvidia.
While Nvidia occupies a dominant position in the GPUs used for the general training of AI, many competitors see space for chips expansion designed for more specific applications. Some of Nvidia’s largest competitors in AI Computing, which are also some of their greatest customs, include cloud suppliers such as Google, Microsoft and Amazon, all of which are building their own custom processors.
NVLINK “Consolida Nvidia as the center of next-generation AI factories, even when those systems are not completely with Nvidia chips,” Wang said, and pointed out that he opens opportunities for Nvidia Nvidia-Banging bugs, Getstems Whostems. Seeking to integrate some of its GPU.
“If it is widely adopted, Nvlink Fusion could expand the footprint of the NVIDIA industry by promoting a deeper collaboration with custom CPU developers and ASIC designers in the construction of the infrastructure of the future of the future,” Wang said.
However, Nvlink Fusion runs the risk of losing the demand of the NVIDIA CPU by allowing Nvidia customers to use alternatives, according to Rolf Bulk, capital research analyst at New Street Research.
Neverberness, “at the system level, additional flexibility improves the competitiveness of NVIDIA GPU -based solutions versus alternative emerging architectures, helping Nvidia maintain their position in the center of AI computing,” he said.
The competitors of Nvidia Broadcom, AMD and Intel are so far absent from the Nvlink fusion ecosystem.
Other updates
Huang opened his opening speech with an update on the next generation of Nvidia Grace Blackwell systems for AI workloads. The company’s “GB300”, which will be launched in the third quarter of this year, will offer a higher general performance in the system, he said.
On Monday, Nvidia also announced the new NVIDIA DGX Cloud Lepton, an AI platform with a computing market that Nvidia said they connect to the world’s developers with tens of thousands or GPUs of a global network or cloud suppliers.
“DGX Cloud Lepton helps address the critical challenge to ensure reliable and high -performance GPU resources by unifying access to AI services in the cloud and GPU’s ability throughout the NVIDIA calculation ecosystem,” said the company in a PRET statement.
In his speech, Huang also announced plans for a new office in Taiwan, where he will also build an AI supercomputer project with Taiwan Foxconn, officially known as Hon Hai Technology Group, the largest electronic manufacturer in the world.
“We are delighted to associate with Foxconn and Taiwan to help build the Taiwan’s infrastructure, and support TSMC and other main companions to advance innovation in the AI and Robotics era,” Huang said.
