Jensen Huang, co-founder and Chief Executive Officer of Nvidia Corp., speaks during the Computex conference in Taipei, Taiwan, on Monday 19 May 2025.
Bloomberg | Bloomberg | Getty images
Nvidia CEO Jensen Huang made a whole series of annovations on Monday and revised new products that are aimed at the center of artificial development and computing of artificial Intellignce.
One of the most striking announcements was a new “Nvlink Fusion” program, with which customers and partners cannot use to use non-Non-Nvidia Central Processing units and graphic processing units together with Nvidia and Nvlink products.
Until now, Nvlink was closed for chips made by Nvidia. NVLink is a technology developed by NVIDIA to connect and exchange data between its GPUs and CPUs.
“NV Link Fusion is so that you can build Semi-Custom AI infrastructure, not only Semi-Custom Chips,” said Huang on the Computex 2025 in Taiwan, the largest electronics conference in Asia.
According to Huang, NVLink-Fusion AI infrastructures allows NVIDIA processors to combine with different CPUs and application-specific integrated circules (ASICs). “In any case, you have the advantage of using the NV Link -infrastructure and the NV Link -Ecosystem.”
Nvidia cheered on Monday that AI -chipping -partners for Nvlink -Fusion already include MediaTek, Marvell” Alchip” Astera Labs” Synopsys And Cadence. Under Nvlink -Fusion, Nvidia -customers love Fujitsu And Qualcomm technologies It will also be Lable to connect their own CPUs from third parties with the GPUs of Nvidia in AI Data Centers, it added.
According to Ray Wang, the NVlink of the Washington-based semiconductor and technology analyst represents Nvidia’s plans to record the share of data centers on the basis of Asic’s, which are traditionally seen as Nvidia matches.
While Nvidia holds a dominant position in GPUs used for general AI training, many competitors see room for expansion in chips that are designed for more specific applications. Sub of Nvidia’s Larganst – competitors in AI Computing – which are also sub of its greatest customers – include cloud providers such as Google, Microsoft and Amazon, all of which are customized Processors from Esir.
Nvlink Fusion “consolidates Nvidia as the center of the next generation AI factories-even when those systems are fully built with NVIDIA chips,” Wang said, and noted that it is open for Nvidia to serve customers who are not nvidia-based systems, but want to be subs. Sub of the GPUs.
“If it is hired a lot, NVLink -Fusion could broaden the sector footprint of Nvidia by building collaboration with adapted CPU developers and ASIC designers when building the AI infrastructure of the future,” Wang said.
NVlink -Fusion, however, runs the risk of reducing NVIDIA CPU by enabling NVIDIA customers to use alternatives, according to Rolf Bulk, a stock investigation analyzes a New Street Research.
Nevertheless, “at system level, the added flexibility improves the competitiveness of NVIDia’s on GPU -based solutions versus alternative emerging architectures, allowing Nvidia to maintain its position in the center of AI Computing,” he said.
Nvidia’s competitors Broadcom, AMD and Intel are so far Abbom of the NVLink Fusion Ecosystem.
Other updates
Huang opened his keynote speech with an update on Nvidia’s next generation Grace Blackwell systems for AI-Deskloads. The “GB300” of the company, which will be released in the third quarter of this year, will offer a higher performance of the system system everywhere, he said.
On Monday, NVIDIA also announced the new Nvidia DGX Cloud Lepton, an AI platform with an calculation marketplace that Nvidia said will connect the AI developers of the world with tens of thousands of GPUs of a global network of cloud providers.
“DGX Cloud Lepton helps to tackle the crucial challenge of reliable, powerful GPU sources for security by gaining access to cloud AI services and GPU capacity in the Nvidia Compute Ecosystem,” the company said in a press release.
In his speech, Huang also announced plans for a new office in Taiwan, where it will build an AI -Supercomputer project with Taiwan’s Foxconn, officially known as Hon Hi Technology Group, the world’s largest electronic production.
“We are delighted to work with Foxconn and Taiwan to help build the AI infrastructure of Taiwan and to support TSMC and other leading companies in promoting innovation in the era of AI and robotics,” said Huang.
