In early 2026, a DNA storage startup called Atlas Biosciences (verify exact company name before publishing) announced a partnership with imec, one of the world’s leading semiconductor research organizations, to bring synthetic DNA storage out of the research lab and into the data center.
The target product, the company’s flagship DNA storage product (exact product name to be verified from the official announcement), stores 13 terabytes of data in a single drop of water, roughly 0.05 cubic centimeters. For context, that’s about the size of a period at the end of this sentence.
The world has a storage problem it doesn’t talk about enough. By some estimates, By some estimates, humanity now generates more data in a matter of days than existed in all of recorded history just two decades ago. Magnetic tape, the technology still quietly running most of the world’s archival storage, is running into physical limits.
LTO-9 cartridges, among the highest-capacity tape formats currently in wide deployment (verify whether LTO-10 has reached broad commercial availability), are approaching the ceiling of what conventional magnetic coatings can hold. The data keeps growing. The tape is not keeping pace.
DNA doesn’t have that problem. One gram of synthetic DNA can theoretically encode hundreds of petabytes of digital data, a density orders of magnitude greater than conventional magnetic tape (specific multiplier should be verified against current LTO capacity specifications). It doesn’t degrade in a decade. It doesn’t require power to maintain. In the right conditions, it can remain stable for thousands of years. The information encoded in ancient mammoth bones and Neanderthal teeth has already proved the principle.
The Last Objection Just Fell
For years, the field had one serious unresolved problem: you could write to DNA, but you couldn’t erase and rewrite it. Every storage medium that couldn’t be overwritten was, by definition, a one-trick archive. Not a hard drive. Not a working part of a modern data infrastructure.
That changed in March 2026, when researchers published a study in a peer-reviewed journal (verify institution, journal name, and publication date from the original paper) demonstrating the first rewritable DNA hard drive.
The technique uses something called a novel encoding and nanopore-based decoding technique (exact technical terminology should be verified against the published paper), a method that allows data to be erased and overwritten, the same way a magnetic drive does. It is, in the dry language of research papers, a significant result. In plain English: the last major objection just disappeared.
Which is exactly the kind of thing that tends to bring commercial money off the sidelines.
The Cost Problem Is Real, But It’s Shrinking
There is still a gap to close. Synthesizing DNA for data storage remains significantly more expensive than magnetic tape, with costs that have been falling but are still orders of magnitude higher per gigabyte. Magnetic tape runs pennies per gigabyte. That’s not a rounding error, it’s two or three orders of magnitude. No enterprise IT department is switching to DNA storage at those economics.
This is where imec’s involvement matters. The organization brings expertise in CMOS ASIC integration, the chip fabrication methodology that drove the cost of semiconductor memory down from dollars to fractions of a cent over the past fifty years.
The imec-Atlas partnership is specifically aimed at applying that same manufacturing logic to DNA synthesis. If the analogy holds, and it won’t hold perfectly, but it rarely needs to, the economics of DNA storage could look very different in five years than they do today.
Atlas has stated it intends to launch commercial DNA storage solutions in the near term (verify stated timeline against official company announcements). That’s an aggressive schedule. But the pieces are in place in a way they weren’t even eighteen months ago: a rewritable encoding method, a semiconductor manufacturing partner, a named product with a named density target.
What Changes If This Works
The obvious beneficiary is long-term archival storage, the cold data that organizations are legally or practically required to keep for decades but rarely need to touch. Government records, genomic databases, financial archives, film libraries. DNA storage is dense, stable, and passive. It fits that use case almost too well.
The less obvious implication is physical. A data center that currently requires a climate-controlled warehouse of spinning drives and tape robots could, in principle, store equivalent information in something closer to a refrigerator. Or a test tube rack.
That’s the part that tends to make people stop and recalculate their assumptions about what “infrastructure” actually means.
This article was researched, written, and edited by our human editorial team. AI tools were used in a limited research-assistant capacity. All claims were independently verified.
