How Much Data Can DNA Actually Store?

DNA encodes information using four bases — A, C, G, and T — which means each base can in principle carry up to 2 bits, and the molecule packs those bases at molecular density. That combination makes DNA, in theory, the densest data-storage medium anyone knows of. The DNA information density record tracks two very different numbers that shouldn’t be conflated: a theoretical raw ceiling near 455 exabytes per gram, and what’s actually been demonstrated with a real, working, error-corrected encoding scheme.

Theory versus a working system

The theoretical figure assumes every base stores its full 2 bits with no redundancy at all — a limit, not a design. Real systems can’t operate anywhere near it, because DNA synthesis and sequencing both introduce errors, and a storage scheme with zero redundancy has no way to detect or correct a single flipped base. Closing that gap is entirely a question of coding theory: how much redundancy is the minimum needed to make the data recoverable in practice.

DNA Fountain

In 2017, Yaniv Erlich and Dina Zielinski published DNA Fountain, an encoding scheme built on fountain codes (the same family of erasure-correcting codes used in some real-world data broadcasting) adapted to DNA’s specific constraints — avoiding repetitive sequences and extreme GC-content that synthesis and sequencing handle poorly. Their demonstrated density of 215 petabytes per gram came with a working proof of concept: they encoded a full computer operating system, a computer virus, a 1895 film, and other files into synthesized DNA and recovered every byte perfectly after sequencing.

Why the gap is still enormous

215 petabytes per gram sounds enormous next to any conventional storage medium, and it is — but it’s still roughly three orders of magnitude below the theoretical ceiling. That gap reflects a genuinely hard practical constraint, not sloppiness: DNA synthesis and sequencing costs and error rates are the real bottleneck, not the coding theory, so closing the gap further depends on progress in wet-lab chemistry as much as in algorithms.

Why it’s here

DNA storage is a clean example of the Registry’s distinction between a theoretical bound and an actually-achieved record — tracked as two separate, clearly labeled numbers rather than one blended figure that would misrepresent either the theory or the engineering.

Primary source

Erlich & Zielinski, “DNA Fountain enables a robust and efficient storage architecture,” Science (2017) ↗