Intel Chief Executive Lip-Bu Tan warned that AI-driven shortages of memory and CPUs could worsen next year as demand for high-bandwidth memory strains supplies of conventional DRAM. Speaking at the AI Infrastructure Summit in Santa Clara, California, Tan said some memory prices had risen five to sevenfold and that companies were delaying expansion because of shortages. He also warned that further increases could pressure low-margin smartphones and laptops, where memory can account for 70%-80% of product costs. Tan urged semiconductor customers to diversify production beyond Taiwan Semiconductor Manufacturing Co., which accounted for more than 70% of global foundry revenue in the first quarter, and said Intel could serve only 50% of CPU customers. Intel is pursuing a foundry revival based on advanced process technology, predictable yields and advanced packaging, including EMIB, while Lee Seok-hee oversees packaging, system integration and back-end technology. Tan said future AI infrastructure will combine CPUs, GPUs, NPUs and application-specific accelerators in heterogeneous systems, with large packages integrating processors, HBM and high-speed input/output chips. Intel agreed in April to supply its 14A process to Tesla and SpaceX’s Terafab project, whose first product is due in mid-2028, despite previous yield challenges.