Extropic Aims to Disrupt the Data Center Bonanza
Extrop, a startup developing an exotic new species of computer chip that involves probabilistic bits, has produced its first working hardware along with evidence that more advanced systems can tackle useful tasks in artificial intelligence and scientific research.
The startup's chips work in a fundamentally different way than chips from Nvidia, AMD, and others, and they promise to be thousands of times more energy efficient when scaled up. With AI companies went billions of dollars in building data centersa completely new approach could offer a much less costly alternative to large arrays of conventional chips.
Extropic calls its processors thermodynamic sampling units, or TSUs, as opposed to central processing units (CPUs) or graphics processing units (GPUs). TSUs use silicon components to harness thermodynamic electron fluctuations, and shape them to model possibilities of various complex systems, such as the weather, or AI models that can generate images, text or videos.
The first working Extropic chip has now been shared with a handful of partners, including frontier AI labs, startups working on weather modeling, and representatives of various governments. (Extropic declined to provide names.)
“This allows all kinds of developers to kick the tires,” says Extropic CEO Guillaume Verdon, who rose to fame in the tech world as a colorful and sometimes controversial called online persona Based Beff Jezos and a new techno philosophy known as effective accelerationism or e/acc to establish the startup. Verdon and his co-founder, Trevor McCourt, who is Extropic's CTO, previously worked on quantum computing at Google before pursuing their new computing approach.
One of those who is now testing the new hardware is Johan Mathe, CEO of Atmo, a startup with AI models that can predict with higher resolution than is otherwise possible. The customers include the Ministry of Defence. Mathe says that Extropic's chips should make it possible to calculate the probabilities of different weather conditions much more efficiently.
Extropic has also released software called TRHML that makes it possible to simulate the behavior of an Extropic chip on a GPU. Mathe has used this software as well as the real chip. “I was able to run a few p-bits and see that they were behaving as they should,” says Mathe.
The company's hardware, called XTR-0, consists of a field-programmable gate array (FPGA) chip, which can be reconfigured for different tasks, combined with two of its first probabilist chips, X-0, each containing a handful of p-bits.
Instead of conventional bits that correspond to either a 1 or a 0, the new chip has probabilistic bits, or p-bits, that model uncertainty. Although limited in scale, the new chip shows the potential of the company's new approach.
“We have a machine learning primitive 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.”
