News AI startup Extropic emerges from stealth with superconducting processors it boldly claims will beat GPUs, CPUs, and TPUs

[Admin]

A startup has emerged from stealth mode, heralding its development and fabrication of a prototype superconducting processor for AI. Extropic says it can create AI accelerators that are 'many orders of magnitude faster and more energy efficient than digital processors (CPUs/GPUs/TPUs/FPGAs).'

AI startup Extropic emerges from stealth with superconducting processors it boldly claims will beat GPUs, CPUs, and TPUs : Read more

 

[vanadiel007]

I feel a song coming up: another one bites the dust from Queen.
I just knew we were back on another round of super conductor articles.

 

[thisisaname]

Hype ready for the next funding around, $14.1 million is next to nothing when it comes to chips.

 

[JTWrenn]

Welcome to the new tech bubble. Tons of money will be poured onto all of these ideas. Loads of investment capital will go into finding the next Nvidia. 99% will all fail. Everything will stay about the same with a few new startups will get purchased by the big players.

The descriptions of this alone sound so tech bubble bs hyped that I wouldn't believe a word they say.

 

[Tbonius]

Theranos?

 

[Rakanyshu]

I'll get a turtleneck and promise that with the "magic of technology" every dollar you send me will be multiplied... all this over hyped startups remind me of the ministers making the big bucks... no one is making you pay for it anyway.

 

[Geef]

I think the guy doing this is the third half cousin to the maker of Star Citizen!

 

[das_stig]

Extropic argues that currently available digital processors are not a good match for AI acceleration. The paper introduces a novel probabilistic paradigm of computing that is said to be distanced from increasingly complex pristine digital computers in favor of something more 'biological' and 'noisy'. Extropic's prototype passive thermodynamic chips run the type of probabilistic algorithms used in AI physically, as a rapid and energy-efficient physics-based process. This makes them much more suitable for current AI algorithms than traditional computing processors, which rather unnaturally try to embrace probability and uncertainty - introducing inefficiency.


and at this point my brain rebooted due to sound bite overload, give me the money pitch.

 

[Diogene7]

There is research in the emerging field of probabilistic computing (p-bit computing). One other option is to use spintronics related technologies, especially technologies derived from Non-Volatile-Memory (NVM) MRAM.

MRAM use Magnetic Tunnel Junctions (MTJ), and one option to create p-bit computing is to tune MRAM MTJ to use their stochastic behaviour : sMTJ.

I think that spintronics related technologies are a growing field, and can easierly be integrated with CMOS technology, and open plenty new opportunities : IMO it is clearly the next step after silicon CMOS technology (beyond CMOS technology).

Much, much more funding for the CHIPS Act should be allocated to kick-start spintronics High Volume Manufacturing (HVM)