
AdWorld Businesses for Sale
Visit World Businesses for Sale
Thinking of selling your business? Buy or sell a business locally, nationally or worldwide. 10% off with code AWN10.
The full story behind my Short: A DARPA-funded University of Michigan project, M3BOLT, layers a PMN-PT crystal under an iron-gallium film so magnetic and plain sound waves act as 1s and 0s. The team projects 100–1,000× faster, more efficient computing. Early design concept.
What if a chip used magnetic and sound waves instead of moving electrons? University of Michigan engineers, with up to $6 million from DARPA's Fast and Curious programme and partners at the University of Wisconsin–Madison, are developing M3BOLT, a transistor alternative. Today's transistors are expected to reach their maximum energy efficiency by 2037, and at a few nanometres electrons leak through even when switched off. M3BOLT layers a ferroelectric PMN-PT crystal under a magnetic iron-gallium film: switching the crystal with an electric field makes the film's magnetic poles wobble, sending a magnetic wave (a magnon); flip the poles and the wobble stops, leaving a plain sound wave. Those two states can be a 1 and a 0. The waves could run at terahertz frequencies, hundreds of times faster than today's gigahertz consumer processors, and the film keeps its state with the power off, so it could double as memory. The team says it could be 100 to 1,000 times faster and more energy efficient. Caveat: these are projections for an early design concept; the devices are still being built (Michigan Engineering, 20 Oct 2026).
Computers have mostly become more efficient by shrinking transistors and packing them closer together. But at a few nanometres, quantum tunnelling lets electrons leak through switched-off transistors, wasting energy, and transistors are expected to hit their maximum energy efficiency by 2037 while demand from AI and edge computing keeps rising [1].
M3BOLT uses two layered materials developed in John Heron's lab: a ferroelectric crystal, lead magnesium niobate-lead titanate (PMN-PT), with a film of magnetic iron-gallium alloy on top. Switching the PMN-PT with an electric field shifts its atoms, which makes the magnetic poles in the iron-gallium wobble in a wave called a magnon. Magnons travel through the film with very little resistance, so less energy becomes heat. Flipping the orientation of the poles stops the wobble, and the waves become simple sound waves; magnetic versus non-magnetic waves serve as the 1 and 0, instead of two voltage levels [1].
Because the waves can move at terahertz frequencies, they could let computers process data hundreds of times faster than today's gigahertz consumer processors. The iron-gallium also keeps its magnetic orientation when powered off, so the same material can act as memory, cutting data shuffling between memory and processor. Roozbeh Tabrizian leads the device design, Wei Lu the circuits, John Heron is principal investigator, and Jiamian Hu's University of Wisconsin team provides physics-based simulations [1].
“1 terahertz = 1,000 gigahertz” is a unit conversion I added. The 100–1,000× figures are the researchers' projections for an early design concept, not measurements. The device sketches are my own, drawn from the release's description of the early design concept. The Michigan Engineering page blocked automated downloads, so the iron-gallium photo comes from the same release as reposted on Nanowerk (credited to Marcin Szczepanski, University of Michigan Engineering). Other images are illustrative, from Wikimedia Commons.
The narration is an AI clone of my own voice (Chatterbox), used with my consent.
What if a chip ran on magnetic waves and sound instead of electrons, and could be up to 1,000 times more efficient? It's called M3BOLT: a University of Michigan project with up to 6 million dollars from DARPA. The problem: today's transistors are expected to hit their efficiency limit by 2037. So the team stacks a crystal called PMN-PT under a thin magnetic film of iron and gallium. Flip the crystal with an electric field, and the film's magnetic poles wobble, sending out a magnetic wave. Flip the magnet, and the wobble stops: just a sound wave. Magnetic wave or plain sound, that's your one and zero. And here's where it gets interesting: those waves could run at terahertz, hundreds of times faster than gigahertz laptop chips. The film even remembers its state with the power off, so it could double as memory. One caveat: this is an early design concept, and the devices are still being built. Why it matters: one day, a big data centre might not need its own nuclear power plant. For the full story, follow the link in the description.
Photos: University of Michigan Engineering (press photo) and Wikimedia Commons (illustrative): 486 Wafer.jpg (Bill Smith, CC BY 2.0); Iron-gallium alloy sample (University of Michigan Engineering press photo) (Marcin Szczepanski, University of Michigan Engineering, press/article photo, credited); Two silicon transistor dies.jpg (Mister rf, CC BY-SA 4.0); PLD Plume.png (H.Perowne, CC BY-SA 3.0); BalticServers data center.jpg (BalticServers.com, CC BY-SA 3.0). Sketches and animation: my own. Narration: AI clone of Nasper Mushonga's own voice (Chatterbox TTS by Resemble AI, MIT licence), used with his consent.

AdWorld Businesses for Sale
Thinking of selling your business? Buy or sell a business locally, nationally or worldwide. 10% off with code AWN10.

AdSharper Image
US$79.99 (USD)
Charge your phone and stand it up hands-free. US store, ships from the US.
Shop now#Ad / affiliate link, I may earn commission.