DNA data storage
Also known as DNA digital data storage, molecular data storage, DNA archiving
DNA data storage encodes digital files in synthetic DNA sequences, then reads them with sequencing. DNA is extremely dense and can remain stable for long periods when stored well, but writing and reading remain slower and more expensive than electronic storage. In 2026, researchers reported advances in composite-letter coding and a codec tested in vitro and in living cells. These studies address parts of the system, not the cost and throughput of a complete archival service.
In plain language
Computer files are long strings of ones and zeros. DNA is also a long code, written with four chemical letters: A, C, G and T. By translating the ones and zeros into these letters, scientists can make short pieces of DNA that hold a file, keep them in a tube, and later read them back with a DNA sequencer.
DNA is attractive because it packs a huge amount of information into a tiny space and can last a very long time if kept cool, dry and protected. The main problems are that writing DNA is slow and expensive, so today it is a research technology aimed at long-term archives rather than everyday storage.
Two 2026 studies show progress in encoding and decoding. Ge and colleagues’ DNA diamond used mixtures of the four natural DNA bases as composite letters. They reported error-free recovery from an eight-letter alphabet at 14-fold sequencing coverage and from a 15-letter alphabet at 33-fold coverage, using pools of 10,000 composite strands. Their cost comparison was modelled using assumed synthesis and sequencing prices, not quoted for a commercial service. Wang and colleagues’ DNA-SaM combined encoding, biochemical sequence checks and a two-tier indexing design. The paper reports successful storage tests in vitro and in living cells; its very large capacity figure is a theoretical estimate based on the code’s sequence design space. Neither result removes the need to improve the cost, speed and reliability of synthesis, physical storage, retrieval and sequencing as an integrated workflow.
Technical explanation
A DNA storage pipeline has four stages: encoding, writing, storing and reading. Encoding maps binary data to base sequences while avoiding error-prone patterns such as long runs of one base, and adds error-correcting redundancy. Writing usually uses array-based synthesis of many short oligonucleotides, each carrying a data payload plus an address. Reading uses high-throughput sequencing, followed by decoding that tolerates missing strands, substitutions, insertions and deletions.
Early demonstrations by Church and colleagues (2012) and Goldman and colleagues (2013) stored a few megabits and recovered them fully. Later work improved robustness and density: Grass and colleagues encapsulated DNA in silica with error-correcting codes, DNA Fountain reached a reported 215 petabytes per gram, and Organick and colleagues stored over 200 MB in more than 13 million oligonucleotides with primer-based random access to individual files. Takahashi and colleagues automated the full write, store and read cycle for a small 5-byte message.
The main bottleneck is writing. A 2022 review put writing speeds in the kilobytes per second range and estimated costs of about 800 million US dollars per terabyte with current technology. Alternatives to base-by-base synthesis are being explored, such as stamping epigenetic methylation marks onto premade DNA templates, which wrote about 275,000 bits without new DNA synthesis in a 2024 Nature study.
What the evidence says
Each statement is typed by the strength of its evidence and linked to the source that supports it.
Demonstrated Shown in a published experiment, test or shipping product.
A 5.27 megabit book was encoded in DNA, written using DNA microchips and read back with next-generation sequencing.
Sources: [1] · AbstractComputer files totalling 739 kB were encoded in synthetic DNA, sequenced and reconstructed with 100% accuracy.
Sources: [2] · AbstractDNA encapsulated in silica, with error-correcting codes, held 83 kB of data and was recovered without errors after one week at 70 degrees C.
Sources: [3] · AbstractThe DNA Fountain scheme stored about 2.14 million bytes, including an operating system and a film, and reported perfect retrieval at a density of 215 petabytes per gram of DNA.
Sources: [4] · Abstract35 files totalling over 200 MB were stored in more than 13 million oligonucleotides, and each file was retrieved individually without errors using random access.
Sources: [5] · AbstractA fully automated device wrote, stored and read back 5 bytes of data, using a custom synthesiser and nanopore sequencing.
Sources: [6] · AbstractEpigenetic movable-type printing wrote about 275,000 bits onto premade DNA templates without de novo synthesis, at 350 bits per reaction on an automated platform, read out by nanopore sequencing.
Sources: [8] · AbstractDNA diamond used composite letters made from mixtures of natural DNA bases. In tests with 10,000-strand pools, its eight-letter system achieved error-free recovery at 14× sequencing coverage and its 15-letter system at 33× coverage. The paper’s cost savings are modelled using assumed synthesis and sequencing unit prices.
Sources: [9] · Abstract and sections on large-scale validation and cost analysisDNA-SaM combines encoding, biochemical sequence constraints and two-tier indexing. The authors report successful storage demonstrations in vitro and in living cells; the paper’s capacity above 2^70 petabytes is a theoretical estimate, not a physically stored dataset.
Sources: [10] · Abstract and discussionReported, unverified Stated by a company or team but not independently confirmed.
A 2022 review, citing earlier estimates, put DNA storage costs at about 800 million US dollars per terabyte with current technologies and current writing speeds in the order of kilobytes per second.
Sources: [7] · Main text, discussion of cost and writing speed (citing earlier estimates) · medium confidenceProposed A design or plan that has not yet been built or tested.
The authors state that their one-week, 70 degrees C test is thermally equivalent to about 2,000 years of storage in central Europe, and that accelerated ageing indicates silica-encapsulated DNA could hold data for millennia under a wide range of conditions.
Sources: [3] · Abstract · medium confidenceLimitations and open problems
- Writing DNA is slow, with a 2022 review placing current speeds in the kilobytes per second range.
- Cost is very high; the same review estimated about 800 million US dollars per terabyte with current technologies.
- Reading is much slower than silicon storage and needs sequencing plus heavy computational decoding.
- Synthesis and sequencing introduce errors and strand dropouts, so substantial redundancy and error correction are needed.
- Fully automated end-to-end systems have so far handled only very small amounts of data, such as 5 bytes.
- The 2026 DNA diamond study used synthesized pools of 10,000 composite strands and demonstrated error-free recovery at 14× coverage for its eight-letter scheme and 33× for its 15-letter scheme. These are research sequencing coverage levels, not system-wide benchmarks for commercial storage.
- The DNA diamond paper’s cost reductions are modelled from assumed unit prices for synthesis and sequencing. The authors identify synthesis as the dominant cost in their analysis, so the estimates should not be read as measured market prices.
- The 2026 DNA-SaM paper reports 100% successful recovery in its in vitro and in vivo demonstrations. Its capacity exceeding 2^70 petabytes is a theoretical estimate, not an experimentally stored amount of data.
Possible future uses
Ideas, not predictions. None of these has been demonstrated.
- Cold archives for data centres that store rarely accessed data for centuries with little energy for upkeep.
- Cultural and scientific heritage archives kept in silica-protected DNA in vaults.
- Hybrid systems where data is searched or processed with molecular operations without first converting it back to electronic form.
- Tagging physical products with embedded DNA records of their origin.
Product concepts: Sealed DNA archive cartridge
Timeline
- Aug 2012Church and colleagues encode a 5.27 megabit book in DNA and read it back by next-generation sequencing. Sources: [1]
- Jan 2013EMBL-EBI researchers store 739 kB of files in synthetic DNA and recover them with 100% accuracy. Sources: [2]
- Mar 2017DNA Fountain coding scheme reports perfect retrieval at 215 petabytes per gram. Sources: [4]
- Feb 2018University of Washington and Microsoft Research demonstrate random access across more than 200 MB stored in DNA. Sources: [5]
- Mar 2019A fully automated end-to-end DNA data storage cycle (write, store and read) is demonstrated for a 5-byte message. Sources: [6]
- Oct 2024Epigenetic movable-type printing writes data onto premade DNA templates without de novo synthesis. Sources: [8]
- Jan 2026DNA diamond reports composite-letter DNA storage with error-free recovery at 14× coverage for an eight-letter alphabet and 33× for a 15-letter alphabet, validated using 10,000-strand pools. Sources: [9]
- 19 Oct 2026DNA-SaM reports a codec and two-tier indexing design tested for data storage in vitro and in living cells. Sources: [10]
Futures this could help bring about
Related breakthroughs
Glossary
- Oligonucleotide
- A short, synthetic strand of DNA, typically tens to a few hundred bases long.
- Sequencing
- Reading the order of bases in DNA molecules, used here to read stored data back.
- Error-correcting code
- Extra, structured information added to data so that errors or missing pieces can be detected and repaired.
- Random access
- Retrieving one chosen file without reading everything else in the archive.
- De novo synthesis
- Building a DNA strand base by base from scratch, rather than copying or modifying existing DNA.
Sources
- peer reviewed paper
Next-generation digital information storage in DNA
George M. Church, Yuan Gao, Sriram Kosuri. Science, 16 Aug 2012. DOI: 10.1126/science.1226355. Accessed 20 Oct 2026. - peer reviewed paper
Towards practical, high-capacity, low-maintenance information storage in synthesized DNA
Goldman N, Bertone P, Chen S et al.. Nature, 23 Jan 2013. DOI: 10.1038/nature11875. Accessed 20 Oct 2026. - peer reviewed paper
Robust chemical preservation of digital information on DNA in silica with error-correcting codes
Robert N. Grass, Reinhard Heckel, Michela Puddu et al.. Angewandte Chemie International Edition, 4 Feb 2015. DOI: 10.1002/anie.201411378. Accessed 20 Oct 2026. - peer reviewed paper
DNA Fountain enables a robust and efficient storage architecture
Yaniv Erlich, Dina Zielinski. Science, Mar 2017. DOI: 10.1126/science.aaj2038. Accessed 20 Oct 2026. - peer reviewed paper
Random access in large-scale DNA data storage
Organick L, Ang SD, Chen YJ et al.. Nature Biotechnology, 19 Feb 2018. DOI: 10.1038/nbt.4079. Accessed 20 Oct 2026. - peer reviewed paper
Demonstration of End-to-End Automation of DNA Data Storage
Christopher N. Takahashi, Bichlien H. Nguyen, Karin Strauss et al.. Scientific Reports, 21 Mar 2019. DOI: 10.1038/s41598-019-41228-8. Accessed 20 Oct 2026. - 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. DOI: 10.1021/acsnano.2c06748. Accessed 20 Oct 2026. - peer reviewed paper
Parallel molecular data storage by printing epigenetic bits on DNA
Cheng Zhang, Ranfeng Wu, Fajia Sun et al.. Nature, 23 Oct 2024. DOI: 10.1038/s41586-024-08040-5. Accessed 20 Oct 2026. - peer reviewed paper
DNA diamond formulates a decomposable composite letter constellation model for DNA data storage
Qi Ge, Menghui Ren, Tingting Qi et al.. Nature Communications, 31 Jan 2026. DOI: 10.1038/s41467-026-68861-y. Accessed 22 Oct 2026. - peer reviewed paper
Towards scalable DNA data storage with DNA-SaM codec system
Yu Wang, Jiaxin Xu, Yang Wang et al.. Nature Communications, 19 Oct 2026. DOI: 10.1038/s41467-026-78759-4. Accessed 22 Oct 2026.
Revision history
4 changes since this profile was created.