Harvard Scientists Build a Silicon Chip That Writes DNA Using Electricity
Researchers at Harvard have built a silicon chip that can write 64 distinct DNA sequences at the same time, using electric currents and water-based enzymes instead of the toxic chemical solvents that conventional DNA manufacturing has relied on for decades. The study was published in Nature Electronics in July 2026, led by Donhee Ham at Harvard's John A. Paulson School of Engineering and Applied Sciences.
The chip does not replace existing methods today. But it points toward a future where synthetic DNA is cheaper, cleaner, and potentially producible closer to the point of use, rather than through centralized industrial facilities.
Why DNA synthesis matters in medicine
Synthetic DNA sits at the foundation of modern medicine's most promising treatments. Gene therapy requires custom DNA sequences. So do personalized vaccines, diagnostic tools, and cancer research applications. The standard method for making this DNA uses phosphoramidite chemistry, which works well but involves hazardous organic solvents and large-scale setups that limit who can produce DNA, how quickly, and at what cost.
Enzymatic synthesis in water has been seen as a cleaner alternative for years, but it has struggled to match the output of the chemical method. Before this work, enzymatic approaches could produce around a dozen sequences simultaneously. Harvard's chip produced 64.
DNA synthesis currently costs roughly $0.05 to $0.10 per nucleotide through commercial providers, a cost that compounds quickly for longer sequences. Gene therapies on the U.S. market today run from hundreds of thousands of dollars to more than $3 million per treatment. A real reduction in manufacturing cost would not automatically lower patient prices, but it does remove a significant barrier in the supply chain.
How the chip actually works
The chip has 256 ring-electrode pairs, each functioning as an independent synthesis site. When a specific site is activated, the inner electrode generates protons that lower the local pH, creating conditions where a water-based enzyme can add a nucleotide to a DNA strand. The outer electrode simultaneously pulls protons away from neighboring sites, keeping the acidic zone from spreading and interfering with adjacent reactions.
Each sequence is assembled one nucleotide at a time, with a blocking group removed between each addition before the next nucleotide can attach. The chip produced sequences up to 39 nucleotides long. As a demonstration, the team encoded 169 bytes of text into DNA, illustrating potential applications in data storage as well as medicine.
The chip started life with a different purpose entirely: recording electrical activity from large populations of neurons. The research team realized the same precision current injection technology could be redirected from cells to molecules. "We wondered whether that same current control could be redirected from cells to molecules," said Ham. "It worked."
What still needs solving
The current bottleneck is not the chip. It is the chemistry. When blocking groups are removed during synthesis, intermediate molecules can drift to adjacent sites and disrupt neighboring reactions. Co-first author Han Sae Jung described it plainly: "The chip did what we asked it to do. The limitation came from the deprotection chemistry, not from the silicon." That leaves a clear next step for the field: develop more localized acid-driven deprotection chemistry that keeps pace with the chip's electronic precision.
The sequences the chip currently produces are also shorter than what most gene therapy or vaccine applications require. Scaling from 64 sequences to millions, and from 39 nucleotides to hundreds, will need both chemistry improvements and new chip architectures. The team estimates five to fifteen years of additional development before clinical applications become realistic.
The bigger picture
Silicon chips have spent half a century powering computing. They are increasingly being used to read and manipulate biology at scale, recording from neurons, reading DNA, and now writing it. This chip is an early but real step toward treating DNA not just as genetic material but as a writable medium that electronics can directly control.
The collaboration involved Harvard, the Broad Institute, DNA Script, and POSTECH. Intellectual property has been filed through Harvard's Office of Technology Development. The paper is available in Nature Electronics under DOI 10.1038/s41928-026-01662-9.