Recently, in a fireside conversation hosted by CASPA on October 24, Intel chief executive Lip-Bu Tan described a two-front agenda: advance the company’s most advanced manufacturing nodes, win CPU performance leadership, and examine whether a competitor, AMD, could become a partner in selected areas.
Tan said Intel has given close attention to both 18A and 14A since he took the helm. He stated that 18A has entered mass production and that the next node, 14A, is also progressing quickly.
He described a PDK path for 14A: the current 0.5 version is available, a 0.9 version is expected soon, and a 1.0 version is planned for early next year to support full production.
In foundry practice, PDK stands for process design kit, a package of design rules, models and technical documents given to chip designers. An early release, such as a 0.5 version, generally represents a testing stage. A 0.9 release is commonly viewed as evidence that the node is ready for tape-out and risk production, marking a major step toward mass production.
Tan also shared an example from his customer-facing work. He said a client waited six to eight months, hoping he would assume Intel’s CEO role. After a formal meeting, the client listed 15 problems from Intel’s past. According to Tan, Intel’s former management sometimes lectured customers instead of listening. He said he addressed the issues one by one, earning the customer’s confidence and leading that client to deepen cooperation with Intel.
On the competitive landscape, Tan confirmed conversations with AMD chairman and CEO Lisa Su about areas where the two companies might cooperate in the future. He did not give details. The disclosure has drawn industry attention because Intel and AMD have long been rivals, and partial cooperation could become an important development in semiconductors.
A recent KeyBanc channel survey reported that Intel has captured foundry orders that include manufacturing network switch chips for Broadcom and server CPUs for Google. The survey also said Intel is expected to produce server processors for Microsoft and gaming GPUs for Nvidia.
It added that Amazon, Anthropic and SpaceX/Tesla Terafab remain prospective customers evaluating Intel’s 18A-P and 14A technologies, with many showing greater interest in 14A.
KeyBanc further indicated that Apple had reached a long-term, multi-generation foundry agreement with Intel covering production of M-series and A-series chips and planned to use 14A rather than the previously rumored 18A-P. The report noted that neither Apple nor Intel has officially confirmed the claim, and some observers still disagree with it.
KeyBanc said yield for substrates used in Intel’s EMIB-T advanced packaging technology is currently around 50% to 60%. Major supplier Ibiden has performed relatively well, while another supplier, UMTC, has held yields in the 20% to 30% range.
As a strategic matter, 18A’s move into volume production, 14A’s approach to tape-out capability, and wider evaluation of Intel’s foundry services by major technology firms show Intel trying to rebuild its position in advanced manufacturing. Tan’s willingness to explore selected cooperation with AMD also opens new scenarios for the industry’s future structure.