A Chip That Writes DNA With Electricity and Water

Every vaccine, every gene therapy, every engineered cell starts with somebody writing a piece of DNA to order. We have been doing that with harsh chemistry since the 1980s. A Harvard team just moved it onto a silicon chip and swapped the solvents for electricity and water.

Here is a thing almost nobody outside a lab thinks about.

Before there is a vaccine, before there is a gene therapy, before there is an engineered cell or a diagnostic test or a synthetic organism, someone has to write a piece of DNA. Not read it. Write it. Build a specific string of genetic letters that did not exist a moment ago, to order, on purpose.

And the way we do that has barely changed since the early 1980s. You place an order. Somewhere, a machine pushes aggressive solvents through a column, bolting on one chemical letter at a time, generating hazardous waste as it goes. It works. It is also slow, dirty, expensive, and stuck inside specialist facilities.

That bottleneck sits underneath the entire field. Which is why what a Harvard-led team just did is bigger than it sounds.

Writing in Nature Electronics, a team led by Donhee Ham, working with the Broad Institute, DNA Script and POSTECH, turned an ordinary silicon chip into a DNA printer.

The chip carries 64 individual synthesis sites on its surface. It wrote 64 different DNA sequences at the same time, in parallel, each up to 39 genetic letters long. And it did it in water, using enzymes, with no harsh solvents anywhere.

For scale: enzyme-based DNA writing had previously topped out at roughly a dozen sequences at once. Sixty-four is not a tweak. It is a different machine.

The elegant part is the control mechanism. Nature already has a tool for building DNA: enzymes, the same molecular machinery your own cells use. The problem has always been telling an enzyme exactly where and when to add a letter, without shouting at the whole beaker at once.

The chip solves that with electricity:

In other words: a semiconductor chip, the same kind of thing that runs your laptop, is now conducting a molecular orchestra with tiny pulses of current. The instruction set is electrical. The output is biological.

It is tempting to file this under "clever lab trick." Do not. The reason it matters is the same reason the printing press mattered, and it has nothing to do with the elegance of the machine.

When the cost and difficulty of writing something collapses, the number of people who get to write explodes. Custom DNA is currently a service you buy from a facility, with a wait and a price tag attached. Both of those quietly decide who gets to experiment. A graduate student with a strange idea and a small budget does not order much DNA.

Put the writing on a chip, and the plausible end state is a DNA printer sitting on an ordinary bench, in an ordinary lab, in an ordinary country. Design in the morning, print by lunch, test in the afternoon. That is the loop that turned software into an industry, and it is precisely the loop biology has never had.

There is an environmental dividend too, and it is not trivial. Swapping a solvent-heavy chemical process for one that runs on enzymes in water removes a genuinely dirty step from the foundation of an entire field.

There is a reason the DNA-writing industry is currently a small number of large, licensed facilities, and it is not only economics.

When you order custom DNA today, the company you order from screens what you asked for. Reputable synthesis providers check sequences against lists of dangerous agents and toxins, and they check who is ordering. It is an imperfect system, but it is a system, and it works precisely because the writing is centralised. A chokepoint you can watch is a chokepoint you can guard.

Now distribute the capability. Put a DNA printer on every bench, in every lab, in every country, and the chokepoint dissolves. That is the whole point of the technology, and it is also the whole problem with it. The same collapse in cost that lets a graduate student test a wild idea also erodes the screening layer that quietly sits between synthetic biology and its worst possible uses.

This is not an argument against the chip. It is an argument for building the screening into the printer before the printer is everywhere, rather than after. The research community knows this. Whether policy moves at the speed of the hardware is a different question.

To prove the point, the team did something slightly show-offy. They took the 64 sequences the chip had written and used them to encode a 169-byte text message, storing digital information inside the DNA itself.

This is a nod at the other prize: DNA data storage. DNA is an absurdly dense way to keep information, and it lasts for thousands of years without electricity, which is more than can be said for any hard drive you have ever owned. The thing that has always blocked it is the cost of writing. Which is, of course, exactly the bottleneck this chip is aimed at.

A 169-byte message is a couple of sentences. Nobody is backing up a data centre in a test tube. But it plants a flag.

The two numbers worth tracking from here are unglamorous and decisive: cost per DNA letter, and maximum strand length. Everything else is commentary.

DNA synthesis means building a specific sequence of genetic letters from scratch, to order, rather than copying one that already exists in nature. It is the starting point for almost all modern biotechnology: vaccines, gene therapies, engineered cells, diagnostics and synthetic biology all begin with a custom-written piece of DNA. Today it is done with a decades-old chemical process that uses harsh solvents and runs in specialist facilities.

The Harvard-led chip has 64 separate sites, each with its own electrode. A precisely controlled electric current changes the acidity at one site, which switches the DNA-building enzymes there on or off. By cycling through the sites, the chip grows 64 different DNA sequences in parallel, in water, using enzymes instead of the usual solvent-heavy chemistry. Enzymatic methods previously managed only about a dozen sequences at a time.

Yes, and the team demonstrated it by encoding a 169-byte text message into the DNA their chip produced. DNA is extraordinarily dense and can survive for thousands of years without power, which makes it attractive for long-term archives. The obstacle has always been the cost of writing DNA, which is exactly the bottleneck this chip attacks. It remains a proof of principle: a 169-byte message is a couple of sentences, not a backup drive.

Related from Frontier Signal: yesterday's deep dive on the AI that reads the words you are about to type. Frontier Signal explains frontier technology in plain English. This is general information, not medical advice.