Next-Gen Storage: Understanding DNA Data Storage and the Future of Archives

We produce more data every two years than in all of human history prior. And yet our storage media — hard drives, SSDs, magnetic tape — all degrade on timescales of years to decades. DNA has stored biological information for millions of years. The question researchers have been working on since 2012: can we deliberately encode arbitrary digital data into synthetic DNA and read it back reliably?

The short answer is yes. The longer answer involves a lot of nuance about cost, speed, and what "practical" actually means.

How It Works

DNA uses four bases: A, T, C, G. By mapping binary data (0s and 1s) to sequences of these bases, researchers can encode digital files as synthetic DNA strands. A gram of DNA can theoretically store around 215 petabytes of data — more than a million consumer hard drives.

The encoding process converts binary to a DNA sequence, then a DNA synthesizer creates the physical strands. Retrieval works in reverse: sequence the DNA, convert back to binary. Error-correcting codes (similar to those used in digital storage) handle synthesis and sequencing errors.

Where It Is Right Now

In 2023, Microsoft and University of Washington demonstrated a fully automated DNA storage system that could write, store, and retrieve data without human intervention. The round-trip time was about 21 hours and cost roughly $1,000 per megabyte — not practical for everyday use, but proof of concept for archival use cases.

By 2025, synthesis costs had dropped roughly 10,000-fold from 2012 levels and were continuing to fall. The trajectory matters more than the absolute number today.

What It's Actually Good For

DNA storage is not competing with SSDs or even tape for hot or warm data. It's targeting the "cold" archival tier: regulatory records, scientific datasets, cultural heritage preservation, anything that needs to survive 100+ years.

The National Archives alone is responsible for petabytes of historical records. Current tape storage requires active maintenance, climate control, and periodic media migration every 10-15 years. DNA stored in a cool, dry environment degrades on geological timescales.

The Barriers That Remain

Speed is the biggest one. Sequencing DNA to retrieve data isn't instantaneous — you're measuring in hours, not milliseconds. For cold archival this is acceptable; for anything with latency requirements, it's not.

Cost is the other constraint, though it's on a rapid downward trajectory. The commercial viability point — where DNA storage competes economically with cloud cold storage tiers like AWS Glacier — is estimated by researchers at somewhere between 2030 and 2035, depending on synthesis cost curves.

The technology is real, the trajectory is promising, and the applications (long-lived archival at extreme density) are genuine. It's not coming for your SSD, but it might eventually replace the magnetic tape libraries in data centers storing things no one looks at but everyone needs to keep.