TLDR
- Intel shares rose Wednesday after DigiTimes reported Nvidia will manufacture parts of its 2028 GPU at Intel facilities
- Nvidia’s Feynman GPU will use Intel’s 18A or 14A processes for input/output die components
- TSMC will build 75% of the chip while Intel handles 25% of packaging and specialized components
- Apple joins the collaboration aimed at meeting U.S. manufacturing requirements
- Nvidia bought $5 billion in Intel stock last December as part of broader support efforts
Intel stock advanced in early Wednesday trading following reports the company secured a manufacturing contract with Nvidia.
DigiTimes said Nvidia will use Intel’s foundry services for its Feynman GPU. The chip launches in 2028 as the successor to Nvidia’s Rubin architecture.
The partnership includes Apple. All three companies are working on low-volume production runs for select components.
This approach satisfies U.S. semiconductor manufacturing requirements. It also maintains their existing relationships with Taiwan Semiconductor Manufacturing Co.
Supply chain sources told DigiTimes that Nvidia will produce specific portions of the Feynman GPU at Intel facilities. The focus is on non-core, lower-tier production runs.
Intel Foundry Wins Key Components
Intel will manufacture the input/output die for the Feynman GPU. This chiplet manages data flow between processor cores and external systems.
The production will leverage Intel’s 18A or 14A manufacturing nodes. These processes represent the company’s cutting-edge foundry capabilities.
Intel will also complete up to 25% of final packaging work. TSMC handles the other 75% of packaging operations.
The core GPU silicon remains with TSMC. Their A16 process node will build components representing about 75% of the chip’s value.
Intel’s EMIB technology will bond the chiplets together. This interconnect system links different components into a single package.
Final assembly takes place on American soil at Intel facilities. This addresses domestic production mandates for certain markets.
Partnership Offers Relief for Troubled Chipmaker
The manufacturing deal arrives as Intel faces serious operational challenges. Nvidia purchased up to $5 billion in Intel shares three months ago.
That investment joined funding from SoftBank and the U.S. government. These backers have provided critical financial support during a difficult period.
Intel has stumbled through multiple strategic errors in recent years. The company surrendered market share across several key product categories.
Last week Intel issued disappointing revenue and profit forecasts. The company said it struggles to fulfill AI server chip orders while new PC processors squeeze margins.
Advanced Technology Targets AI Workloads
The Feynman GPU will incorporate HBM4e or possibly HBM5 memory. These next-generation memory standards enable higher bandwidth and capacity.
Increased memory per package allows the chips to run trillion-parameter AI models. This capability targets the most demanding artificial intelligence applications.
Nvidia has not publicly confirmed the partnership. Intel also has not commented on the DigiTimes report.
The deal gives Intel a high-profile customer for its foundry business. The company has been trying to attract external clients to its manufacturing operations.
Intel’s foundry division needs more customers to justify massive capital investments. The company is building new fabrication plants across the United States.
DigiTimes has a mixed track record on supply chain reports. The publication accurately reports some developments while missing on others.
Intel stock had been under pressure before Wednesday’s gains. Shares have declined over the past year as the company works through its turnaround plan.


