Budget smartphone and laptop manufacturers face mounting production risks as AI-driven demand pushes memory costs higher, Intel Chief Executive Lip-Bu Tan warned. He expects the shortage to worsen next year, with additional manufacturing capacity proving insufficient and memory prices already reaching five to seven times earlier levels.
Tan discussed the problem at the AI Infrastructure Forum in Santa Clara, California. Asked which challenges he wanted prospective AI infrastructure founders to tackle, he highlighted memory availability and recalled that his warning early last year had received limited recognition.
That constraint has since materialized, he said, with restricted capacity and difficulty obtaining sufficient memory already forcing numerous projects to be postponed. Manufacturers trying to build midrange and lower-priced phones or laptops are also finding procurement increasingly difficult.
Behind the disruption is the increasing importance of high-bandwidth memory, or HBM, in AI computing. Processors, including graphics processing units, face limits on how much computing performance they can deliver on their own, increasing the role of memory designed for AI workloads.
Samsung Electronics, SK hynix and Micron have been unable to supply enough HBM to satisfy the jump in demand. Manufacturers’ shift toward HBM production has also constrained the availability of conventional memory, spreading the shortages beyond specialized AI components.
For manufacturers of inexpensive devices, the challenge is particularly acute. Tan said memory now represents between 70% and 80% of the total cost of budget smartphones and laptops, leaving companies struggling to accommodate further increases. Continued price escalation could make these low-margin products uneconomic to manufacture before higher-end equipment faces comparable difficulties.
The exposure includes phones priced below $200, which account for most handset shipments worldwide, alongside basic laptops used by schools, small businesses and price-sensitive consumers. With such a large portion of their costs tied to memory, suppliers have little scope to protect buyers from the increase. The available responses include charging more or reducing RAM and storage compared with the preceding generation.
Premium hardware has more room to withstand the same component inflation as its costs are distributed differently. Camera systems, screens and the brand premium account for substantial spending on a $1,000 smartphone.
The pressure also extends beyond consumer electronics. Nvidia has decided that its AI server products will cost at least 15% more beginning next year, attributing the increase to more expensive memory semiconductors.
