Podcast Episode
Intel will handle up to twenty five percent of the final packaging requirements using its EMIB (Embedded Multi-die Interconnect Bridge) technology, while TSMC will oversee the remaining seventy five percent. The main compute die will continue to be produced by TSMC, preserving Nvidia's relationship with the world's leading contract chipmaker.
For Intel, the collaboration offers validation of its foundry capabilities as the company works to attract external customers under CEO Lip-Bu Tan. Intel's foundry division has faced significant challenges, posting multi-billion dollar operating losses in recent years.
Intel shares rose more than three percent following the report, signalling market optimism about the chipmaker's foundry prospects.
Nvidia Teams Up with Intel for 2028 Feynman GPU Production
January 28, 2026
Audio archived. Episodes older than 60 days are removed to save server storage. Story details remain below.
Nvidia is reportedly partnering with Intel to manufacture key components of its next-generation Feynman GPU architecture, set for 2028. The deal would see Intel produce the I/O die and handle around twenty five percent of packaging, while TSMC continues making the main compute die.
A Historic Semiconductor Partnership Takes Shape
Nvidia is planning to collaborate with Intel on manufacturing portions of its next-generation Feynman GPU architecture, according to a report from Taiwan-based DigiTimes citing supply chain sources. The move represents a significant shift in Nvidia's manufacturing strategy as American tech giants adopt dual-foundry approaches.How the Manufacturing Split Works
Under the reported arrangement, Intel would manufacture the I/O die for the Feynman GPU using its 18A or 14A process nodes, with mass production expected around 2028. The I/O die manages communication between processor cores and external components, representing a non-core but essential part of the chip architecture.Intel will handle up to twenty five percent of the final packaging requirements using its EMIB (Embedded Multi-die Interconnect Bridge) technology, while TSMC will oversee the remaining seventy five percent. The main compute die will continue to be produced by TSMC, preserving Nvidia's relationship with the world's leading contract chipmaker.
Strategic Motivations Behind the Deal
The partnership reflects a broader industry shift toward supply chain diversification under the Trump administration's Made in America objectives and tariff pressures. Tech companies including Apple, Google, Microsoft, and Amazon Web Services are reportedly in similar discussions with Intel.For Intel, the collaboration offers validation of its foundry capabilities as the company works to attract external customers under CEO Lip-Bu Tan. Intel's foundry division has faced significant challenges, posting multi-billion dollar operating losses in recent years.
What is Feynman?
Named after theoretical physicist Richard Feynman, this architecture represents Nvidia's next-generation data centre platform following the Rubin series. First revealed at GTC 2025, Feynman will feature next-generation high bandwidth memory, eighth generation NVSwitch, and is expected to deliver five to twenty times the compute increases over Rubin chips.Intel shares rose more than three percent following the report, signalling market optimism about the chipmaker's foundry prospects.
Published January 28, 2026 at 2:48pm
