Extropic AI
Building thermodynamic computing hardware that is radically more energy-efficient than GPUs by leveraging natural entropy for probabilistic AI workloads.
At a Glance
- AI researchers
- Data center operators
- Semiconductor manufacturers
AI Tools by Extropic AI
(1)
thrml
Open Source Thermodynamic ML Library
Discussions
No discussions yet
Be the first to start a discussion about Extropic AI
Latest News
Inside X0 and XTR-0: Launching our new hardware platforms
How Extropic Plans to Unseat Nvidia
Guillaume Verdon presents at TED AI San Francisco
Extropic Emerges from Stealth with $14.1M Seed Funding
Products & Services
A hardware platform that enables low-latency communication between Extropic's thermodynamic chips and traditional processors.
Probabilistic sampling hardware cores designed to run generative AI workloads with extreme energy efficiency.
An open-source Python library for developing thermodynamic machine learning algorithms and simulating them on TSUs.
Market Position
Positions itself as a physics-first alternative to digital accelerators like NVIDIA GPUs, focusing on energy efficiency and probabilistic alignment.
Leadership
Founders
Guillaume Verdon
Former Research Scientist at Alphabet Inc. (Google Quantum AI and X); Lead of Quantum ML; Founder of TensorFlow Quantum. Known as @basedbeffjezos, a key figure in the effective accelerationism (e/acc) movement.
Trevor McCourt
Former Student Researcher at Google; Background in Physics and Engineering.
Executive Team
Guillaume Verdon
Founder & CEO
Ex-Google Quantum AI, pioneer in quantum machine learning and thermodynamic computing.
Trevor McCourt
Chief Technology Officer
Ex-Google, expert in hardware-software co-design for probabilistic systems.
Board of Directors
Founding Story
Founded by former members of Alphabet's quantum computing research team (Google Quantum AI and X) to solve the energy crisis in AI. The company aims to replace traditional digital logic with physics-based probabilistic computing that works with, rather than against, the natural fluctuations of matter.
Business Model
Revenue Model
Likely B2B hardware sales and platform licensing for its thermodynamic computing hardware and integration platforms.
Pricing Tiers
Hardware development platform for early adopters and researchers.
Target Markets
- AI researchers
- Data center operators
- Semiconductor manufacturers
- Generative AI model inference
- Energy-efficient data center compute
- Research in thermodynamic and probabilistic algorithms
- Limited public info - targeting early deep tech researchers and data centers.