Extropic aims to disrupt the Bonanza data center

external, A startup developing a strange new type of computer chip that handles probabilistic bits has produced its first working device as well as proof that more advanced systems can handle tasks useful in artificial intelligence and scientific research.

The startup’s chips work in a radically different way than chips from Nvidia, AMD and others, and promise to be thousands of times more power efficient when they scale up. With the influx of artificial intelligence companies Billions of dollars to build data centersa completely new approach could offer a much lower cost alternative to large arrays of traditional chips.

Extropic calls its processors thermodynamic sampling units, or TSUs, rather than central processing units (CPUs) or graphics processing units (GPUs). TSUs use silicon components to harness thermodynamic electron fluctuations, shaping them to model the probabilities of various complex systems, such as weather, or artificial intelligence models capable of generating images, text or videos.

The first working Extropic chip has now been shared with a handful of partners, including frontier AI labs, weather modeling startups, and representatives from several governments. (Extropic declined to provide names.)

“This allows all kinds of developers to kick the tires,” says Guillaume Verdon, CEO of Extropic, who has gained notoriety in the tech world as a colorful and sometimes controversial online personality. Based on Beef Jesus A new techno philosophy known as… Effective acceleration Or e/acc before founding the startup. Verdon and co-founder Trevor McCourt, who is CTO at Extropic, previously worked on quantum computing at Google before pursuing their new computational approach.

One of those now testing the new devices is Johan Matthi, CEO of Atmo, a startup that uses artificial intelligence models that can predict with higher accuracy than is otherwise possible. Among its clients is the Ministry of Defense. Extropic chips should make it possible to calculate the probabilities of different weather conditions much more efficiently, Mathie says.

Extropic also releases software called TRHML that makes it possible to emulate the behavior of an Extropic chip on a GPU. Mathe used this software as well as the real chip. “I was able to turn on a few of the p-bits and noticed that they behaved the way they were supposed to behave,” Mathie says.

The company’s hardware, called the XTR-0, consists of a field-programmable gate array (FPGA) chip, which can be reconfigured for different tasks, as well as two of its first probabilistic chips, the

Instead of traditional bits that correspond to a 1 or 0, the new chip features probability bits, or p-bits, which represent typical uncertainty. Although limited in size, the new chip demonstrates the potential of the company’s new approach.

“We have a primitive machine learning system that is much more efficient than matrix multiplication,” McCourt says. “The question is, how do you build something on the scale of ChatGPT or Midjourney.”