TLDR
- Intel stock jumped Wednesday on reports Nvidia will use Intel manufacturing for 2028 Feynman GPU components
- Intel’s 18A or 14A process nodes will produce input/output die sections of the Feynman architecture
- Intel gets 25% of packaging work while TSMC maintains 75% production share
- Deal follows Nvidia’s $5 billion investment in Intel from December 2024
- Partnership with Apple included to satisfy U.S. chipmaking compliance requirements
Intel stock climbed in premarket trading Wednesday after reports emerged that Nvidia selected Intel for future GPU production.
DigiTimes reported Nvidia plans to manufacture portions of its Feynman GPU at Intel foundries. The chip targets a 2028 release date.
The collaboration also involves Apple. The companies focus on low-volume production for non-core chip components.
This strategy meets U.S. semiconductor manufacturing mandates. It preserves working relationships with Taiwan Semiconductor Manufacturing Co.
Supply chain sources said Nvidia will produce specific Feynman GPU sections at Intel facilities. These represent lower-tier production runs focused on specialized components.
Intel Secures Input/Output Die Production
Intel will build the input/output die for Nvidia’s Feynman GPU. This component controls data transfer between processor cores and external devices.
Production uses Intel’s advanced 18A or 14A manufacturing processes. These nodes showcase Intel’s latest foundry technology.
Intel also handles 25% of final packaging operations. TSMC manages the remaining 75% of packaging work.
TSMC still produces the majority of the Feynman chip. Their A16 process node creates components worth approximately 75% of total chip value.
Intel’s EMIB interconnect technology bonds chiplets together. This system integrates separate components into one functional package.
Final assembly occurs at Intel’s U.S.-based facilities. This meets domestic manufacturing requirements for government and enterprise customers.
Deal Provides Support During Difficult Period
The partnership arrives as Intel navigates major business challenges. Nvidia purchased up to $5 billion worth of Intel shares in December.
That investment combined with support from SoftBank and U.S. government programs. Intel needed this financial backing during its restructuring efforts.
Intel lost ground to competitors over the past several years. Strategic missteps damaged the company’s market position and financial health.
The chipmaker released weak quarterly guidance last week. Intel cannot fulfill surging AI server chip demand while new PC processors reduce profit margins.
Feynman Targets Advanced AI Applications
The Feynman GPU incorporates HBM4e or HBM5 memory technology. These next-generation standards deliver higher performance than current memory types.
Greater memory capacity enables trillion-parameter AI model processing. This capability serves the most demanding artificial intelligence workloads.
Nvidia’s current Rubin GPU precedes the Feynman architecture. The new design promises performance gains for AI computing tasks.
Intel’s foundry business needs more external customers. The company invested billions in new U.S. manufacturing facilities.
Landing Nvidia as a customer validates Intel’s foundry strategy. It demonstrates the company can compete for high-profile chip production contracts.
Neither Nvidia nor Intel confirmed the DigiTimes report. Both companies declined to comment on the partnership details.
DigiTimes sources supply chain information from Taiwan. The publication’s accuracy varies across different reports and predictions.
Intel shares had declined over the previous year. The stock suffered as investors questioned the company’s turnaround timeline.
Wednesday’s gains came as investors weighed the potential impact of the Nvidia deal. The partnership could help Intel’s struggling foundry division attract additional clients.
