The method builds on primitives validated in Extropic's X0
Our paper on Denoising Thermodynamic Models is now published in Nature's npj Unconventional Computing.
We demonstrate thermodynamic hardware built on the primitives validated in X0 could unlock generative modelling at 10,000x lower energy than GPUs.
https://www.nature.com/articles/s44335-026-00075-3
Users are excited about Extropic's thermodynamic hardware paper claiming 10,000x energy efficiency gains for generative AI because of the major efficiency leap and its publication in a top journal like Nature.
No Digg Deeper questions have been answered for this story yet.
The stage is set for Z1.
More announcements coming soon.
Our paper on Denoising Thermodynamic Models is now published in Nature's npj Unconventional Computing.
We demonstrate thermodynamic hardware built on the primitives validated in X0 could unlock generative modelling at 10,000x lower energy than GPUs.
https://www.nature.com/articles/s44335-026-00075-3
@beffjezos congrats!!
The stage is set for Z1.
More announcements coming soon.
@willccbb ty will!
@beffjezos @grok what does this mean
@beffjezos @willccbb Fuck ya!! Finally some thermo shit let’s fucking go good work dude
@beffjezos Congrats
@beffjezos 10,000x is unreal
@beffjezos LFG!!!
Beff Jezos (Guillaume Verdon, Extropic founder) is saying their key paper just published in Nature's npj Unconventional Computing.
It shows their thermodynamic/probabilistic hardware (using transistor noise for sampling) can run diffusion-like generative models at ~10,000x lower energy than GPUs on benchmarks.
"The stage is set for Z1" = their next commercial chip (after X0 prototype, with ~250k p-bits) is now validated and incoming, plus more announcements soon.
It's a big proof point for their ultra-efficient AI hardware approach.
@beffjezos Can't wait to see this! 🔥
@extropic What are the key milestones to enable mass manufacturing.
Are there still difficult technical challenge?
10k lower is crazy.
It's a gift from the thermodynamic God hiding in plan sight.
Dans le chaos naît l'opportunité!
@beffjezos @willccbb I still think you should sell test kits, I’d pay a lot
@beffjezos
@beffjezos Published in Nature. A top and well respected journal. This is where things get serious.
Going to have a good read of this paper. Congrats!
@beffjezos Cool!
@beffjezos hey Beff
What about new name for nerfe Claude
Did Feeble won ?
@beffjezos very exciting, beff! may your p(boom) be infinite!🚀 I look forward to one day soon turning one of these OFF and ON again, haha.
@extropic Eh