About TrendForce News

TrendForce News operates independently from our research team, curating key semiconductor and tech updates to support timely, informed decisions.

[News] Mega NVIDIA-Intel Deal: No Chipmaking Orders, Co-Developed Products Might Launch 2027–28


2025-09-19 Semiconductors editor

NVIDIA on Thursday stunned markets with a $5 billion investment in Intel and the announcement of a joint effort to develop chips for PCs and data centers—a striking vote of confidence in the struggling U.S. chipmaker, coming just weeks after Washington took a historic 10% federal stake. In addition, the move will make NVIDIA one of Intel’s largest shareholders with a stake of around 4% once new shares are issued, as per Reuters.

The announcement quickly raised one key question: would Intel’s foundry play a role in manufacturing? According to Reuters and Tom’s Hardware, Intel Foundry will indeed supply the central processors and advanced packaging for the jointly developed products. While the partnership is not a chipmaking deal, the reports note that Intel will provide advanced packaging services—potentially including its Foveros 3D and EMIB technologies.

Notably, The Verge reports CEO Jensen Huang confirmed at the conference call that NVIDIA’s partnership with TSMC remains unchanged, and clarified that the deal isn’t about moving chip manufacturing to the U.S.

“We’re fully committed to the Arm roadmap, we have lots and lots of customers for Arm. Nor is it a shift from TSMC to Intel as manufacturing partner for NVIDIA’s chips,” he said, as per The Verge.

Another key question is when the co-developed products will reach the market. While no official timeline has been given, Tom’s Hardware reports that the joint client project—pairing an Intel CPU with an Nvidia GPU chiplet—is expected to take three to four years from concept to mass production.

In detail, the report suggests that the collaboration demands deep integration across SoC design, performance and power targets, packaging technologies like Foveros and EMIB, and software stacks. Work likely began in 2024, with the first products expected in late 2027 or early 2028, the report adds.

Intel-NVIDIA Ties Brewing for Over a Year

Tom’s Hardware, citing Jensen Huang, highlights that the two parties’ collaboration traces back more than a year, with preliminary agreements struck between then–Intel CEO Pat Gelsinger and Huang even earlier.

As CNBC explains, NVIDIA’s AI systems currently run on Arm-based CPUs. However, under the agreement, the AI chip giant will soon integrate Intel CPUs into its NVLink AI racks. Meanwhile, NVIDIA will also provide GPU technology for Intel chips in laptops and PCs.

Jensen Huang, cited by CNBC, estimated the combined addressable market for these collaborations at $50 billion. As the report notes, the joint projects will also use Intel’s advanced packaging, which integrates multiple chip components into a single part for installation in machines.

This collaboration is highly significant for Intel, signaling its further step into the AI market. According to Reuters, AI servers need to connect dozens—or even hundreds—of chips to handle massive data workloads as a single system. Until now, NVIDIA’s top-selling AI servers with these ultra-fast links relied solely on its own processors. The new pact, however, puts Intel on equal footing, giving the company a share of every NVIDIA server sold, Reuters reports.

Impact on Semiconductor Heavyweights

As Reuters points out, despite Huang’s praise for NVIDIA’s TSMC partnership, the deal still poses risks for the Taiwanese foundry, which currently makes NVIDIA’s flagship processors. As NVIDIA aligns more closely with Intel, some of that lucrative work could eventually move away from TSMC, the report warns.

On the other hand, the report also warns that AMD, a key competitor to Intel in data-center chips, could feel the impact as NVIDIA’s backing bolsters Intel in the fiercely contested server market.

Additionally, the collaboration could also intensify competition with Broadcom, a leader in chip-to-chip connectivity that powers AI efforts at companies like Google, as noted by Reuters.

Read more

(Photo credit: Lip-Bu Tan’s X)

Please note that this article cites information from Reuters, CNBC, Tom’s Hardware and The Verge.


Get in touch with us