Forget magnetic tape, enterprises looking to archive huge quantities of data might be using a method pioneered by life itself: DNA storage.
The technological breakthrough can’t come soon enough. North American data center vacancy rates have been steadily dropping for years according to trends tracked by datacenterHawk, while JLL’s data centers global outlook report for 2024 projected long-term challenges arising from exponential growth in power use and storage requirements.
DNA data storage has one big advantage—it’s incredibly dense. Harvard scientists estimated a single gram of DNA can hold around 215 petabytes of information. It’s also extremely stable, meaning it could outlast mechanical or digital storage methods by decades or more, and it’s not like the format is going to change.
“This standard has been around for billions of years, and is likely to be around for as long as there’s life on Earth,” Rob Carlson, managing director of tech venture capital firm Planetary Technologies and affiliate professor at the University of Washington, told IT Brew. With Gartner projecting enterprise storage shortfalls potentially reaching two-thirds of demand by 2030, Carlson thinks DNA could bridge the gap.
While scientists have been able to sequence DNA for decades, and the creation of synthetic DNA has gotten cheaper, its use in data storage is still limited to niche cases. For example, Wired reported, French company Biomemory began offering DNA storage chips in late 2023 for $1,000 a pair—with a max capacity of a kilobyte.
“A primary challenge for stakeholders in this field is developing a core technology that can efficiently scale to exabytes and zettabytes,” Erfane Arwani, Biomemory’s CEO, told IT Brew via email.
In 2018, Carlson worked with colleagues at UW and Microsoft to build what he calls the “first soup-to-nuts automated prototype,” capable of reading, storing, and writing in DNA. However, that device had a throughput of just five bytes over around 21 hours. Carlson estimated in a recent IEEE Spectrum article current global synthetic DNA capacity would need to grow by around four orders of magnitude to compete with a single two-gigabit-per-second archival tape drive.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
“Who knows exactly what format that will take, the production of the DNA, the construction of the molecules themselves,” Carlson told IT Brew. “The future I see is a semiconductor chip.”
The method the UW/Microsoft team used in 2018 isn’t viable for commercial use in part because it requires large quantities of acetonitrile, a volatile solvent.
One solution, Carlson said, could be enzymes, which biochemists currently use to produce synthetic DNA for purposes like PCR testing. Last year, his team published research alongside Ansa Biotechnologies demonstrating electrochemical control of single-base enzymatic additions.
“We’re talking about many orders of magnitude scale up required to produce enough DNA to store a lot of data in it,” Carlson told IT Brew. “And so I think that we have to have this interface that converts digital signals to chemical signals, and then the means to convert the chemical signals to molecules will be this enzymatic system.”
Arwani and Carlson both agree long-term commercial prospects for DNA storage are contingent on making it accessible to everyday technologists without expertise in biochemistry.
“It is currently unthinkable for data centers to employ biologists or chemists in lab coats operating complex and fragile machines,” Arwani wrote. “It’s essential to have machines that function apparently like current storage arrays, which can be inserted into racks and use the same interfaces as traditional products.”
“If you ask me to explain specifically how the magical terabyte flash drive I carry around works, I'm going to be doing some hand waving,” Carlson said. “It still works.”