Google and NVIDIA are exploring Intel as a potential backup manufacturing partner for advanced AI chips, according to a report by The Information.
The move could strengthen Intel’s ambitions to become a major contract chipmaker while reducing the industry’s dependence on Taiwan Semiconductor Manufacturing Company (TSMC).
Google has reportedly placed an order with Intel to manufacture more than three million of its in-house Tensor Processing Units (TPUs) in 2028.
Google’s commitment represents a concrete step to secure future capacity for its in-house AI accelerators, which power services like Search, cloud computing, and emerging agent platforms. The order would leverage Intel’s 18A process node and advanced packaging technologies, the report added.
NVIDIA has not committed to a production order but is evaluating Intel’s manufacturing technology for future AI processors. The tests reportedly include Intel’s advanced packaging capabilities and 18A process, which could be used to produce complex chips that integrate four GPUs into a single package, including future architectures such as Feynman.
The discussions come as soaring demand for AI hardware continues to strain TSMC’s manufacturing capacity, prompting some of the world’s largest AI chip designers to seek alternative production partners.
Intel’s foundry business, which manufactures chips for external customers, has been a key focus area under CEO Lip-Bu Tan as the company attempts to regain ground lost to TSMC over the past decade.


Your reaction
Discussion
What's your take on this story?
No comments yet — be the first to share your view.