TechNewsReel
Live

Vaire Computing validates reversible chips that recycle 50% of energy

A new test chip proves that 'running calculations backward' can recover waste heat, offering a potential escape from the AI-driven thermal wall.

TechNewsReel Newsroom · September 8, 2026

Vaire Computing has successfully demonstrated a test chip that recycles energy typically lost as waste heat, marking a physical breakthrough for the theoretical field of reversible computing. The development suggests a path toward drastically reducing the energy intensity of data centers and consumer electronics as AI demand surges.

The startup's demo chip, codenamed 'Ice River' and fabricated in 22 nm CMOS, utilizes a patent-pending resonator to achieve net energy recovery. The chip is capable of recycling approximately 50% of its energy intensity on average. Hannah Earley, co-founder and CTO of Vaire Computing, describes the resonator as a "glorified pendulum" designed to store and reuse energy that would otherwise dissipate.

The physics of energy recovery

Conventional computing operates by erasing intermediate information during calculations, a process that inevitably dissipates energy as heat—a phenomenon known as Landauer's principle. Reversible computing avoids this loss by retaining intermediate data, allowing the computational process to be run backward to recover the energy. While the concept was proposed 50 years ago, it remained largely theoretical until Vaire's implementation of adiabatic switching in physical hardware.

Earley argues that this approach represents a fundamental shift in hardware design. "Reversible computing can be so much more energy efficient than conventional computing, and it’s potentially the way we should have originally built computers," Earley told New Scientist. She added that if a chip were designed today from first principles, it would likely follow this reversible architecture.

Breaking the thermal wall

This breakthrough arrives as the industry faces a critical thermal wall. Traditional chip scaling, governed by Moore's Law, is struggling to keep pace with the explosion of AI workloads, which require massive amounts of power and generate immense heat. By decoupling computational power from heat generation, reversible computing could eventually lead to "near-zero energy" chips.

Such a shift would fundamentally alter the economics and environmental impact of global computing infrastructure. Instead of relying on increasingly complex cooling systems to remove waste heat, chips would treat energy as a recyclable resource, potentially lowering the operational costs of the world's largest data centers.

Scaling toward commercialization

To date, Vaire Computing has raised over $12 million in funding to advance this architecture, according to MIT Technology Review. The company is now focused on rebuilding chip architecture from the ground up to move beyond the initial test phase.

While the Ice River chip proves the concept, the path to commercialization requires scaling this technology to complex, general-purpose processors. Observers will be watching to see if Vaire can maintain these recovery percentages as they move to smaller process nodes and more intricate circuit designs.

Sources

Get a notification when a big story breaks. A few a day at most — no spam.