I have been co-chair of a Mass Storage Roadmap effort with Roger Hoyt for over two decades. In 2022 we moved this activity to the IEEE International Roadmap for Devices and Systems (IRDS) roadmap activity. We released the 2026 roadmap recently with additional leadership help from Riyan Mendonsa from Seagate in Minneapolis. The mass storage roadmap committee included representatives from IBM, WD, Kioxia, Seagate, Folio Photonics, Cerabyte, SPhononix and some industry consultants.

Topics covered in the 2026 roadmap included HDD and magnetic tape, solid state storage including NAND flash as well as new memories such as magnetic random-access memory (MRAM), ferroelectric memory (FRAM), resistive ram (RRAM) and phase change memory, (PCM). The optical storage section included roadmaps for Folio Photonics, Cerabyte, SPhotonix and Optera Data. We also had a section focusing on DNA storage. Between the various optical storage and DNA storage activities there is currently a big focus on storage media that can last a long time for archival applications.

The 2026 mass storage roadmap is available here .

The roadmap group had interactions with other roadmap groups within IRDS including the Beyond CMOS group that developed projections on new solid-state memories. Of course, the roadmap took into account the severe shortage of every type of memory and storage due to the growth of demand to support AI data centers.

The figure below, from Coughlin Associates, tracks HDD areal density growth, both in laboratory demonstrations and in shipping products, since 2000.

Energy assisted magnetic recording, in particular HAMR, plus more disks and head will enable HDDs with 100+TB before the end of the decade.

Magnetic tape is led by the LTO format, which projects projects up to 1PB compressed capacity in the 2030’s.

Solid state storage included NAND flash as well as new solid-state non-volatile memories. We used the NAND flash roadmap from the Beyond CMOS IRDS group, which projects over 1,000 layers by the early 2030’s and a 3D bit density at the maximum number of bits/cell of 1.83X10 13 bits/cm 2 . The table below ( from New Memories, Not Just for AI ) shows a comparison of new non-volatile memories with conventional SRAM, DRAM and NOR and NAND flash.

The table below is from the optical storage section and includes conventional Blu-Ray and Archival Discs as well as expected products from Folio Photonics, SPhotonix, Cerabyte and Optera Data. These optical storage startups, focusing on archival storage, include products from Folio Photonics and Optera that would use conventional optical disc formats as well as rectangular storage media from Cerabyte and SPhotonix.

The roadmap also looked at the future of DNA storage which can be extremely dense, with a theoretical raw bit density of 1M TB per cubic millimeter with and a reasonable delivered density of 60TB per cubic millimeter. Under the proper storage conditions (cool temperatures and excluding water and oxygen) it can last a long time.

The biggest challenge for many DNA data approaches today is the slow and expensive write and read speeds. Proponents say that the development of genomic medicine will drive down the costs of writing and reading synthetic DNA, eventually making DNA a cost-effective archival storage media.

The IEEE Mass Storage Roadmap shows a path to digital storage technologies supporting AI workloads to the long-term preservation of human knowledge and experience.