DigitalGenetics - stock.adobe.co
Iridia is working on a way to overcome the cost barrier to using synthetic DNA as a data storage medium.
CEO Murali Prahalad said other companies working on DNA storage, such as Microsoft and Catalog, are too focused on lowering the cost of reading DNA and exploiting the technology to its full potential. Iridia's approach is to instead focus on creating a self-contained chip that will fit in the same form factor as disk and tape today, but at much higher data density. Trade-offs will necessarily be made to balance density and cost, but the idea is to bring to market something that will beat what's available.
The idea attracted investor attention, as earlier this month, Iridia raised $24 million in its Series B funding round. The round was led by LifeSci Venture Partners of New York and included Western Digital Capital, the investment arm of the American hard drive manufacturer.
SearchStorage spoke with Prahalad about how Iridia's technology helps customers, how far along it is and how the company will use its funding to make DNA-based storage viable in the storage market.
What's the current dilemma with synthetic DNA as a data storage medium?
Murali Prahalad: It's a basic math problem. If we believe IDC numbers this year, we're on track to generate something like 60 zettabytes of data globally, and the global average of what gets stored in archives is about 3% of the total. If I were to presume that only 0.1% of that 3% were written to DNA through some form, and then had to be read back via a traditional DNA sequencer, it would be the physical equivalent of sequencing 2.9 billion human genomes.
You physically could not commandeer the entire infrastructure of sequencers in the world to try to do that. And even if you could, at today's sequencing costs, you're talking about a $1.4 trillion exercise, which translates to $1 million cost per terabyte. This knocks it out of any kind of contention from a cost basis compared to what's available in magnetic tape and magnetic hard drives.
What's Iridia's solution to that problem, and how is it different from other vendors exploring DNA storage?
Prahalad: Right now, for data center folks, the whole idea of moving liquids in tubes and sequencers is alien to how their operations run on a day- to-day basis. So for us, it's not just about writing DNA better, faster, cheaper, but putting it in a chip form factor that's portable and interoperable enough that it could make its way to a data center.
We would incorporate it in the form of a chip that has vastly superior read-write latencies than magnetic tape and has power consumptions up to 2,500 times lower than flash, and, if you think about this in the context of an exabyte-level data center, you could potentially reduce operating expenses up to 99%. So the goal is to get to that chip.
The big line of demarcation I can see here is, if you look at all our competitors, every one of them has focused on only one thing: the synthesis of DNA and trying to drive down the cost. For example, Twist Bioscience and Microsoft have a partnership where Twist basically provides a chemical synthesis means, and if you look at other groups, they're trying to do the same. They're all different variations on a theme.
The dirty secret is no one is really confronting the read aspect. The prevailing assumption is that Illumina or some other sequencing company is going to handle the read aspect. I think it's a very skewed view to what the actual solution really needs to be.
What's your plan for bringing this technology to the market?
Prahalad: We're working through the process of developing post-fab filling of the chips. Ultimately, what that means is we will not be at the hypothetical, theoretical max of DNA storage. Because while you can maybe have higher densities, your cost structure gets so ridiculous that no one's going to use it.
We have a reduction in the density, but we still have densities that are maybe four orders of magnitude what's being contemplated in tape or hard drive, but at a cost structure that becomes competitive in terms of per-terabyte versus bit rot.
Murali PrahaladCEO, Iridia
Our goal is to just build the best darn chip we can and get it to the densities that we're targeting. If we get it to something in the form factor and footprint of a hard drive or magnetic tape reel, a data center operator doesn't care if it's filled with a whole bunch of DNA-based chips.
We looked at the roadmaps for tape and hard drives to see what costs we would have to meet or beat to be competitive. We essentially worked out all these costing models, and I think that's what's attracting strategic investors to us. Western Digital, and others in the tape world, they're asking to partner with Iridia, because they're looking at their own internal roadmaps and saying, 'If you're half-right, we've got to pay attention.'
How far along is this chip?
Prahalad: We've shown we can fabricate the [nanopore-based] memory cells, move the data at will and manipulate it. We can write bits to that DNA strand, and we can read the bits we've written electronically. Now it's a function of chewing through the raw error rates. We're already thinking about what's the best in terms of density and signal optimization, and so forth.
We're effectively building the world's first nano-memory cell chip that will be synthesized in a fab and filled with the reagents, and then use CMOS controls to write data to it and read it back. That is then essentially the first warp drive, and if that thing works the way we hope and we get the raw data rates that we're anticipating, then scaling just becomes a linear engineering exercise of jumping the density of nano-memory cells per square centimeter and the number of bits per nano-memory.
We feel all the componentry at the unit level is functioning the way we want to, and now it's the question of building that integrated chip.
What do you plan to do with the $24 million from your recent Series B funding round?
Prahalad: We're hiring. We're just 10 people and we're looking to double to 20. Whether it's experts in fabrication science, surface chemistry, digital signal processing, material science, enzymology -- we're now looking to get some of the best minds out there, and we're now in a great financial position to do that.
And the other part is working with fabs we've selected to actually build out the chip, and then there's a number of other external R&D [projects] that will get done.
Western Digital Capital was one of the investors. Can you tell me more about your relationship with the storage vendor?
Prahalad: We're super appreciative of their investment, but this is not any kind of exclusive relationship. They are a phenomenal partner, and one we're really thrilled to have, but we're really anticipating a global product.
We'll work with Western Digital as this evolves, and over time, we'll see what's the most optimal commercial channel and commercial strategy. Obviously, they'll be at the table, but this is not an exclusive partnership in any respect.
The building of this integrated chip really presents a lot of monetization opportunities, and you can imagine a roadmap of higher and higher density chips. Now, should others come and look at the company and say, 'Hey, this is too irresistible,' we'll just continue to see that as it reveals itself, as opposed to try to engineer a specific outcome.
I think right now, the team is just dedicated on building out the tech and proving it.