Gene therapies can treat cancer, genetic diseases, and antibiotic-resistant infections, but often stumble on a simple problem: the synthetic DNA they work with relies on cumbersome and toxic chemistry. Researchers at Harvard School of Engineering and Applied Sciences led by Professor Donhee Ham published a chip in Nature Electronics in June 2026 that could break this bottleneck. The semiconductor system synthesizes 64 different DNA sequences simultaneously using enzymes and water instead of organic solvents.
What Enzymatic DNA Synthesis Means
Behind every DNA strand manufactured today for a drug or research application lies phosphoramidite chemistry from the 1980s. It works, but has significant drawbacks: reactions require water-free conditions and solvents like acetonitrile, trichloroacetic acid, and pyridine. The oligonucleotide industry produces more than 1.1 million liters of hazardous chemical waste annually with this approach, according to estimates cited multiple times in Nature Methods.
The Harvard chip works differently. It uses the enzyme TdT (terminal deoxynucleotidyl transferase), which can extend DNA strands without a template. The chip contains 64 synthesis sites, each with two concentric ring electrodes. The inner electrode electrolyzes water and locally releases protons, lowering the pH. This precisely activates the enzyme reaction at that site. The outer electrode simultaneously pulls protons away, preventing the reaction from affecting neighboring sites. The result: 64 different sequences grow in parallel without cross-contamination.
The team collaborated with the Broad Institute, South Korean Postech University, and French biotech company DNA Script. DNA Script already distributes a commercial enzymatic synthesizer and has been connected since 2020 to Harvard and Broad through an IARPA-funded program for DNA data storage.
Why Enzymes Outperform Chemicals
Phosphoramidite chemistry has a practical length limit of approximately 200 nucleotides before error rates and yield become too poor. Enzymatic synthesis promises longer strands with lower error rates because TdT operates in mild aqueous conditions. Additionally, reactions can proceed without anhydrous environments, which previously required specialized and expensive laboratory equipment.
DNA Script's commercial Syntax synthesizer already produces up to 120 nucleotides per strand enzymatically. The innovation of Harvard's chip lies in electrical control: instead of physical fluid systems delivering individual nucleotides to synthesis sites, electricity takes over. This makes the technology miniaturizable and scalable on a semiconductor chip.
In Comparison: From Single-Strand Device to 64-Fold Parallel Chip
Enzymatic synthesis is not new. The principle using TdT enzymes has been known since the 1990s. The step toward controlled parallel synthesis on a chip represents the critical advance. A Twist Bioscience team demonstrated in 2023 in Science Advances how to enzymatically control 12 different sequences simultaneously on a chip. The Harvard chip jumps in one step to 64 locations.
For cost comparison: the global DNA synthesis market had a volume of approximately 5.2 billion dollars in 2025, at a price of roughly seven cents per base pair for standard oligonucleotides. A 2021 MIT estimate calculated the cost of storing one petabyte of data in DNA at around one trillion dollars, explaining why cost reductions in this field would have substantial consequences. The industry index synthesis.cc documents: the price per base pair declined from approximately 30 dollars in 1990 to seven cents today, a drop exceeding 99 percent in 35 years.
Three Hurdles Until Gene Therapy for Everyone
The chip produces sequences up to 39 nucleotides in laboratory validation. Gene therapeutic applications typically require sequences of hundreds to thousands of nucleotides. Three development steps separate current demonstration from clinical application.
First, sequence length must increase. TdT can theoretically produce longer strands; whether Harvard's chip's electrical pH control maintains sufficient precision for long strands must be demonstrated. Second, scaling is needed: 64 synthesis sites are a laboratory proof-of-concept; production volumes for therapy development would require thousands of parallel sites. Third, error rates remain in focus: enzymatic methods promise better accuracy than chemical methods, but this must still be demonstrated for longer strands in this chip architecture.
Harvard has filed patents through its Office of Technology Development. DNA Script as commercial partner is well positioned to integrate chip technology into future devices. The team does not name a concrete timeline. Whether the chip accelerates the price decline in DNA synthesis as solar cells have undergone in renewable energy is the truly decisive question for gene therapy accessibility.