Photonics and optical computing · Experimental cross-disciplinary systems
Archival storage: DNA versus glass
How do DNA, laser-written glass and magnetic tape compare for long-term archival storage, on density, lifetime claims, speed and deployment stage?
The 2022 DNA review says DNA storage is not yet competitive with conventional storage, and Microsoft says the glass research phase is complete with no product announced. Each has figures under different test conditions. Magnetic tape is the only option here with a product on sale in the cited sources, and its figures come from a reseller article that does not say where they come from.
| Option | DensityData stored per unit of medium volume (Gbit/mm³) or per cartridge (TB), as reported in the cited work. Blank where no reliable public figure was found. | Lifetime claim and test basisYears or millennia the medium is claimed to keep data readable, with the test or basis stated. Extrapolated lab results are not observed ageing. | Read and write speedWriting and reading throughput in bytes or bits per second, with the scale or basis stated (per beam, per run, or per drive). | Deployment stageWhere the technology sits between laboratory work and commercial sale, as stated in the cited source. |
|---|---|---|---|---|
| DNA data storage | No reliable public figure | Unprotected DNA (in solution or dried): half-life on the order of months to a few years. Inorganic-matrix-encapsulated DNA: millennia, per the review's estimates at 10 °C.Reported, unverified[1] | Writing: in the order of kilobytes per second, per the review's estimate for current archival writing. Reading: much lower than standard silicon-based storage media, per the review.Reported, unverified[1] | The 2022 review states that DNA data storage, at present, is not competitive with magnetic and solid-state storage devices.Reported, unverified[1] |
| Laser-written glass | 1.59 Gbit/mm³ for fused silica (birefringent voxels, 301 layers, 120 mm square, 2 mm platter). 0.678 Gbit/mm³ for borosilicate (phase voxels).Demonstrated[2] | Above 10,000 years at 290 °C for borosilicate phase voxels, extrapolated from Arrhenius fits to 440–500 °C accelerated tests. Thermal erasure only; mechanical stress and chemical corrosion not tested.Reported, unverified[2][3] | Writing: 25.6 Mbit/s peak per beam for fused silica, limited by the 10 MHz laser repetition rate (516 nm). The paper does not report read throughput.Demonstrated[2] | Microsoft's 2026 blog says the research phase is complete. It announces no commercial product or release date.Reported, unverified[3] |
| Magnetic tape | No reliable public figure | Shelf life of 30+ years, stated in a reseller article with no storage conditions.Reported, unverified[5] | LTO-10: 400 MB/s native transfer rate, as listed in a reseller article.Reported, unverified[5] | Commercial product series. A 2025 reseller article reports the LTO-10 launch, and the LTO Program site says LTO is currently in its 10th generation.Reported, unverified[5][4] |
What the table shows
- Each row uses a different test basis, so the values sit side by side without a common measure.
- Laser-written glass has a density and a peak per-beam write speed from a 2026 paper. The DNA write speed is an estimate, in the order of kilobytes per second, from a 2022 review.
- Magnetic tape is the only row with a commercial product in the cited sources. The DNA review says DNA storage is not yet competitive, and Microsoft says the glass research phase is complete.
- The DNA and glass lifetime figures rest on accelerated tests, extrapolation or estimates. The tape figure is a shelf-life statement with no storage conditions.
Read with care
- DNA density is left blank because none of the cited sources gives a figure that can be checked against the others.
- Glass density figures are reported for specific samples and voxel designs, not as a product specification.
- Lifetime figures are not comparable: glass values are extrapolations from thermal tests only, DNA values are estimates from lab work, and tape values are shelf-life statements with no conditions.
- Tape density is left blank because the cited sources give cartridge capacity but not volumetric density, so the three rows cannot be compared on that measure.
Sources
- review article
Emerging Approaches to DNA Data Storage: Challenges and Prospects
Andrea Doricchi, Casey M. Platnich, Andreas Gimpel et al.. ACS Nano, 18 Oct 2022. Accessed 21 Oct 2026. - peer reviewed paper
Laser writing in glass for dense, fast and efficient archival data storage
Microsoft Research Project Silica Team. Nature, 18 Feb 2026. Accessed 21 Oct 2026. - company announcement
Project Silica's advances in glass storage technology
Richard Black. Microsoft Research, 18 Feb 2026. Accessed 21 Oct 2026. - standard
LTO Roadmap
LTO Program, 2026. Accessed 21 Oct 2026. - news
LTO-10 launches with breakthrough 30TB capacity: a game changer for archival storage
CXP, 1 Aug 2025. Accessed 21 Oct 2026.