Summit Explores Post-Von Neumann Computing From Superconductors to Biology

Data centers are hitting a physical ceiling, forcing researchers to abandon traditional chip architecture in favor of hardware that operates at the absolute limits of energy efficiency.

At the recent Imagination in Action Minds in Motion summit in Boston, innovators unveiled designs that move beyond the current constraints of transistor scaling and Von Neumann systems.

Jake Read detailed a shift where machines function as feedback devices that understand their own physics. By merging CNC operations with machine code, these systems enable higher-order functions.

Eric Strand argued that moving data carries a high energy cost. His research focuses on locality and specialization, demonstrating that reconfigurable volume mount devices and RISC-V systems-on-chip can outperform hardware like the H200 in core arithmetic operations, Forbes and tech.yahoo.com reported.

The urgency stems from the physical limits of current hardware. Camron Blackburn said that data center power consumption is reaching a breaking point as transistors hit their minimum possible size.

Photo: tech.yahoo.com

“We’re reaching a point where transistors can’t keep shrinking,” said Camron Blackburn.

To bypass these limitations, researchers are looking toward Landauer’s limit, which suggests a fundamental minimum energy requirement for switching states that is 100,000 times lower than current levels. Blackburn is currently focusing on superconducting devices to bridge this gap.

“We really need superconducting devices in order to fill this gap, from where transistors operate today, to that fundamental minimum,” said Blackburn.

Unlike CMOS devices that require energy to pull voltage levels up or down, AQFP devices encode information in circulating currents, allowing them to function at their potential energy minimum.

Neil Gershenfeld highlighted the vast efficiency gap between engineered neural nets and biological systems, which operate at eight orders of magnitude greater efficiency. He proposed shifting from a model where a computer controls a machine to one where the computer itself is the machine.

“In biology, the key insight is: the code doesn’t describe the thing; the code becomes the thing,” said Neil Gershenfeld.

Neil Gershenfeld, a professor in the MIT Program in Media Arts and Sciences and director of the MIT Center for Bits and Atoms, specializes in interdisciplinary research spanning quantum computing, nanotechnology, and personal fabrication. As an intellectual leader in these fields, he pioneered the 1998 development of the first 2-qubit quantum computer capable of processing data and outputting solutions.

His work at the Center for Bits and Atoms focuses on merging digital fabrication with computing to enable the creation of personalized technology. In 2013, he and his team introduced reversibly assembled cellular composite materials, advancing the integration of structural hierarchy into engineered systems. By fostering the Fab lab movement, he continues to explore how digital fabrication can empower individuals to create sophisticated machines on demand.

This approach seeks to recreate the structural hierarchy of molecular biology within engineered materials. While the practical implementation of these cryogenic racks and bio-inspired architectures remains under development, the research points toward a fundamental departure from the Von Neumann architecture that has defined modern computing.

Photo of author

Technology Desk

Technology Desk at Archyde.

Some Players Find Vice City Least Interesting Part of GTA VI