MW Nvidia just made this 'very smart move' that could pay off big-time down the road

By Britney Nguyen

The company will allow data centers to use custom silicon from other chip makers with Nvidia's AI infrastructure

Nvidia Corp.'s new technology will make it easier for customers to use chips from other companies - and that could be a good thing for Nvidia's business down the road, according to analysts.

The chip maker's NVLink Fusion, unveiled during its GTC Taipei event in Taiwan on Monday, will allow customers to integrate custom silicon from other companies with Nvidia's $(NVDA)$ artificial-intelligence infrastructure, such as its graphics processing units, in data centers for training and inferencing. Nvidia developed the NVLink technology to connect its GPUs and central processing units to share data between them.

Making room for other chip makers is "a wise move as the AI [data center] market is still very concentrated among a few players," which have all designed their own AI chips and have different infrastructure requirements, Richard Windsor, founder of research firm Radio Free Mobile, said in a note on Monday. And the opening "is unlikely to hurt Nvidia's GPU market share or its margins," he added.

MediaTek Inc. (TW:2454), Marvell Technology Inc. $(MRVL)$ and Synopsys Inc. $(SNPS)$ are among Nvidia's partners developing custom silicon that is compatible with NVLink Fusion, the company said at its event ahead of Computex 2025. And CPUs from Fujitsu Ltd. (JP:6702) and Qualcomm Inc. $(QCOM)$ will be able to integrate with the chip maker's GPUs in data centers, Nvidia added.

"A tectonic shift is underway: for the first time in decades, data centers must be fundamentally rearchitected - AI is being fused into every computing platform," Nvidia Chief Executive Jensen Huang said in a statement. "NVLink Fusion opens NVIDIA's AI platform and rich ecosystem for partners to build specialized AI infrastructures."

The new silicon will make it "very, very hard for a hyperscaler not to buy a lot from [Nvidia] even if it uses some custom silicon (CPU or XPU) since they now offer an infrastructure design on a single Nvidia networking fabric," Melius Research analysts said in a note on Monday. NVLink Fusion "is a very smart move that keeps Nvidia at the center of your AI Data Center wallet," the analysts said, because it makes it possible to scale up custom silicon to meet increasing demand for AI. The analysts added that they think it "hints at" Nvidia's intention to eventually sell more software solutions.

Bank of America said in a note on Monday that NVLink is an opportunity for Nvidia to expand its total addressable market, because it "opens up [Nvidia's] proprietary interconnect and networking tech" to custom accelerators and CPUs from other companies.

Nvidia also introduced its RTX PRO servers that are built with its Blackwell server-edition GPUs, which can run legacy applications and handle enterprise-size AI workloads, while also enabling multimodal AI inferencing and physical AI. The new enterprise servers are "driving the shift from CPU-based systems to efficient GPU-accelerated infrastructure," Nvidia said.

"This is yet another threat for Intel $(INTC)$, whose [data-center] business is already under siege," Windsor said. Now Intel "will now have to fight yet another defensive battle."

He continued: "How much success Nvidia will see with this is unclear at this stage, but if it can do a good job of migrating the legacy to accelerated GPUs cost-effectively, then this is a good proposition for any enterprise looking to start using AI in its systems."

Meanwhile, the RTX PRO servers "could really help drive enterprise AI adoption in the U.S. since it is looking increasingly likely that we could see legislation that allows for 100% depreciation write offs for U.S. Capex," the Melius analysts wrote.

Windsor noted that Nvidia is looking at "another large revenue opportunity" with its RTX PRO servers, and that interest is likely to be high due to Intel's faltering business. Pushing a shift to GPUs from CPUs also poses a threat to Intel's legacy x86 architecture, he added.

While Nvidia won't see "immediate financial benefit" from either of these announcements, "in the long run, they will serve to support Nvidia's market share and augment revenue and profit growth," Windsor said.

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

May 19, 2025 13:18 ET (17:18 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.