On October 19, 2020, Intel sold its NAND flash and storage business to SK Hynix for $9 billion. This included Intel's SSD business, NAND components and wafer operations, and its NAND flash chip manufacturing plant in Dalian, although Intel retained its Optane business. After SK Hynix made the final payment to Intel in March 2025, these assets were transferred to its newly established subsidiary, Solidigm. Meanwhile, Intel exited the storage market.
According to TrendForce, Intel CEO Lip-Bu Tan recently stated that new memory architectures, once considered commoditized, have now become strategically significant. The memory industry is entering a mature phase of innovation and has become one of Intel's key areas of focus. While people may already guess what he is contemplating, he is not yet ready to elaborate.
According to Lip-Bu Tan, he was previously a proponent of the view that one should "not invest in the memory industry," but the situation has now changed. While Tan did not disclose specific ideas or Intel's plans, he believes that stacking memory directly onto CPUs is "very reasonable." There are many ways to stack the two together, which may be the direction Intel will explore for its future new architectures.
Although Intel is now widely known as a semiconductor company primarily focused on developing and manufacturing CPUs, it originally built its business on memory products. It wasn't until the 1980s, when intense competition from Japanese manufacturers caused significant losses in its memory division, that Intel completely shifted its focus to CPUs. Later, Intel attempted to re-enter the market with NAND flash and Optane technology, but ultimately withdrew again due to strategic downsizing.
Earlier this year, Intel announced a collaboration with Powerchip Semiconductor Manufacturing Corporation (PSMC) and SoftBank subsidiary SAIMEMORY to develop a new memory technology called "Z-Angle Memory (ZAM)." Last month, Intel published a new patent regarding its XBM memory, which is considered a practical application of ZAM and an alternative to HBM4, with commercialization expected around 2030.