AI & Technology

Synthetic DNA Integrated into Semiconductor Chips to Deliver 100x Lower-Power AI Memory Architecture

In a breakthrough published in Advanced Functional Materials, Penn State engineers merge synthetic DNA with crystalline perovskite to build bio-hybrid memristors that process and store data in the same location.

By 19Network Editorial Team · Aug 17, 2026 · 5 min read

Microscopic strands of synthetic DNA weave through a dark semiconductor chip to form a glowing bio-hybrid memory circuit.

Engineers develop a bio-hybrid memristor combining synthetic DNA with perovskite semiconductors, slashing AI memory power consumption by up to 100 times.

Key Takeaways

Bio-Hybrid Breakthrough: Researchers at Penn State University integrated chemically engineered synthetic DNA with crystalline perovskite semiconductors to create high-density, ultra-low-power memory devices.
x Energy Efficiency: The resulting memory resistor (memristor) architecture consumes up to 100 times less electrical energy than conventional solid-state storage devices while operating at less than 0.1 volt.
In-Memory Neuromorphic Computing: Stores and processes computational data in the exact same physical space, mimicking the biological synapses of the human brain to eliminate the traditional Von Neumann memory bottleneck.
Thermal & Environmental Stability: Unlike earlier biological computing experiments, the DNA-perovskite device demonstrated thermal stability up to 250°F (121°C) and operated reliably at room temperature for over six weeks.
Frontier AI Scalability: Offers a hardware solution to the escalating energy and cooling demands of enterprise artificial intelligence data centers and edge neural processors.

Materials scientists and computational engineers have taken a historic step toward resolving the escalating energy crisis in artificial intelligence by successfully merging biological molecules with solid-state electronics. In a study published in the peer-reviewed journal Advanced Functional Materials , a research team led by Pennsylvania State University revealed that combining synthetic DNA with crystalline perovskite semiconductors creates a bio-hybrid memory architecture that consumes up to 100 times less power than standard silicon storage devices. The technological milestone addresses the primary limitation facing modern computing: the memory bottleneck in artificial intelligence hardware. As large-scale frontier AI models scale toward trillion-parameter architectures, standard computer chips face severe thermodynamic and electrical constraints. Traditional silicon architectures are built on the Von Neumann model, where computational processing (CPU/GPU) and memory storage (DRAM/NAND) are physically separated. Shuttling continuous streams of data back and forth across interconnects generates immense waste heat, consumes gigawatts of electricity, and throttles real-time inference speeds. To overcome this structural barrier, researchers turned to nature’s most dense and durable information storage medium: DNA. In biological organisms, DNA stores approximately 215 million gigabytes of information per gram. However, biological molecules have historically been…

Read on 19Network