This new brain-like chip could slash AI energy use by 70%

2026-04-23 · Source: Robotics Research News -- ScienceDaily · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Engineering & Applied Sciences · Depth: Expert, short

Summary

Researchers at the University of Cambridge have developed a new nanoelectronic device using a modified form of hafnium oxide that mimics neuronal processing, potentially reducing AI energy consumption by up to 70%. Published in *Science Advances*, this brain-inspired chip functions as a stable, low-energy memristor by combining memory and processing. Unlike conventional chips that move data between separate units, this device changes resistance by adjusting energy barriers at p-n junctions within a hafnium-based thin film, rather than relying on unpredictable conductive filaments. This design enables ultra-low switching currents, hundreds of stable conductance levels, and biological learning behaviors like spike-timing dependent plasticity, offering a more efficient alternative to current power-hungry AI hardware.

Key takeaway

For AI Hardware Engineers designing next-generation systems, this hafnium oxide memristor offers a path to drastically reduce energy consumption. Your focus should be on overcoming the current 700°C manufacturing temperature challenge to integrate this technology into practical chip-scale systems, which could enable more adaptable and smarter AI with significantly lower operational costs.

Key insights

A new hafnium oxide memristor design significantly cuts AI energy use by mimicking brain function.

Principles

• Combine memory and processing for efficiency.
• Interface switching offers superior uniformity.
• Ultra-low currents enable energy savings.

Method

Engineered a hafnium-based thin film with strontium and titanium, using a two-step growth process to create p-n junctions at interfaces, allowing resistance changes by adjusting energy barriers.

In practice

• Integrate memristors for energy-efficient AI.
• Explore hafnium oxide for neuromorphic hardware.
• Apply spike-timing plasticity in learning systems.

Topics

• Brain-Inspired Computing
• Neuromorphic Hardware
• Hafnium Oxide Memristors
• AI Energy Efficiency
• P-N Junctions

Best for: AI Scientist, AI Hardware Engineer, Research Scientist

Related on AIssential

• This Chip Breakthrough Shrinks Data Centers 10,000× — Superconducting logic offers ultra-dense, energy-efficient classical computing by eliminating electrical resistance and heat, addressing AI's data movement bottleneck.
• Brain-inspired phototransistor could cut AI energy use by sensing and storing data — Brain-inspired phototransistor integrates sensing and memory to reduce AI energy consumption.
• Indian Researchers Develop Molecular Memristor for Neuromorphic Computing — Molecular memristors offer high-resolution, energy-efficient neuromorphic computing for edge AI by performing in-memory analog computation.
• Scientists built a memory chip that breaks the rules of miniaturization — Extreme miniaturization combined with structural innovation can enhance memory performance, defying conventional scaling limits.
• Princeton builds bio-hybrid computer using living neurons — A 3D bio-hybrid computer integrates living neurons with electronics to create energy-efficient, pattern-recognizing neural networks.
• Otters++: A Time-to-first-spike Based Energy Efficient Optical Spiking Transformer — Otters++ leverages optoelectronic device decay for energy-efficient time-to-first-spike computation in spiking Transformers.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Robotics Research News -- ScienceDaily.