Intel chief executive Lip Bu Tan has opened direct talks with AMD leader Lisa Su regarding potential partnership opportunities between the 2 rival semiconductor makers. The dialogue marks an aggressive shift in corporate strategy as Intel works to rebuild customer confidence and open its fabrication facilities to outside clients. Concrete details of any formal joint pact remain unconfirmed by either board.
Tan acknowledged previous management missteps during a fireside discussion hosted by the Chinese American Semiconductor Professional Association. Intel previously built processors strictly according to internal roadmaps while dismissing client requests for tailored silicon modifications. The company now plans to customize its Xeon server processors to match client workloads while targeting application specific integrated circuits, copackaged optics, and advanced packaging to deliver roughly 10% of total revenue each.
Tan addressed the previous corporate mindset directly.
Intel is so proud they never want to modify for customer needs. I am modifying it for customer needs, and so that, I think, to me, is a fast growing area, and I am double, triple down on it.
Tan noted that 1 prospective customer waited over 6 months to present a list detailing 15 distinct areas where Intel had failed to deliver. The executive reviewed and resolved each critique individually before the client agreed to expand commercial ties with the company.
Intel Foundry operations form the second pillar of the turnaround effort. The 18A manufacturing process has achieved volume production, while the subsequent 14A node is advancing through internal development stages. Intel distributed version 0.5 of the 14A process design kit to chip designers, with version 0.9 scheduled next and volume production version 1.0 arriving in early 2027.
Confidence remains tied to verified performance targets.
We have to reclaim the leadership in CPU. That is something that I am very determined to do.
Bilateral collaboration between the 2 competitors is already materializing in software engineering. AMD agreed to merge its ZenDNN optimizations directly into the Intel open source oneDNN deep learning library, governed under the Unified Acceleration Foundation. This integration provides developers with a single software pipeline for PyTorch artificial intelligence workloads across both AMD and Intel CPUs, removing fragmentation across the x86 ecosystem.
The engineering teams are simultaneously drafting AI Compute Extensions to speed up matrix multiplication on x86 hardware. Intel enterprise program manager Jeff McVeigh highlighted the developer benefits of the shared library.
Developers should not have to think about separate software paths to get the best from an x86 CPU. Working with AMD through oneDNN gives the PyTorch community a consistent approach for modern processors and a way to bring future capabilities into the framework.
AMD architecture corporate vice president Robert Hormuth pointed to the broader standard setting effort.
By working with the x86 Ecosystem Advisory Group, we are bringing the industry together around common technologies that simplify development across x86 processors.
