Big Idea:

Extropic, a hardware startup founded by former Alphabet quantum research team members, has secured $14.1 million in seed funding. Extropic is reportedly developing a chip optimized for running LLMs. The technology is described as a "novel full-stack paradigm of physics-based computing" that incorporates concepts from non-equilibrium thermodynamics, hinting at a unique approach to chip design.

Why It Matters:

Traditional computing models are constrained by physical and complexity limitations. Extropic's breakthrough aims to build a system where computing errors or noise, which has been a significant challenge in quantum computing, are considered an asset rather than a liability. The company has not disclosed detailed technical information but indicated a goal to reduce the electricity required for running AI models and automate certain coding tasks, potentially allowing the computer to self-program for learning representations of the world.

Key Details:

• Founded in 2022 by CEO Guillaume Verdon and CTO Trevor McCourt, both of whom had previously led a quantum computing team at Alphabet's X research unit. They were involved in developing a TensorFlow library for running AI models on quantum computing chips.
• Funding round led by Kindred Ventures, with participation from HOF Capital, Julian Capital, Marque VC, OSS Capital, Valor Equity Partners, Weekend Fund, and executives from Adobe Inc., Shopify Inc., and other AI startups.

What's Next:

Extropic is advancing the development of their physics-based computing model, which could automate coding tasks and enhance processing of LLMs, potentially leading to major improvements in AI applications.

Dive Deeper:

Contrary to its founders' background in quantum computing, Extropic's product is not a quantum computing chip. The decision is attributed to the scalability challenges of quantum physics-based computers and the company's pursuit of a different path to practical physics-based computing.

Zoom Out:

This development challenges Nvidia's dominance in AI hardware, especially with its H200 data center processor. Extropic's approach, utilizing physics-based computing, offers a compelling alternative in the competitive AI chip industry.