Start your day with intelligence. Get The OODA Daily Pulse.

Home > Companies > Quantum > Extropic

Extropic

Not quantum computing, Thermodynamic Computing. But certainly influenced by their founder’s deep knowledge of how things work at the quantum scale.

Extropic is a California-based deep tech startup developing a revolutionary new class of AI hardware—thermodynamic computing chips—designed to radically improve the efficiency and capability of generative AI and large language models (LLMs). Founded in 2022 by quantum computing pioneer Guillaume Verdon, Extropic’s mission is to merge the power of generative AI with the physical world by leveraging the natural properties of thermodynamics and stochastic processes. The company’s technology promises to extend Moore’s Law, enabling AI systems that are orders of magnitude faster and more energy-efficient than today’s digital hardware, with broad implications for national security, edge computing, and the future of artificial intelligence.

Leadership

  • Guillaume Verdon (Founder & CEO):
    Verdon is a physicist and quantum computing researcher recognized for pioneering quantum deep learning and co-founding Google’s TensorFlow Quantum project. Before Extropic, he served as quantum tech lead at Alphabet’s X and contributed to Google Quantum AI. He holds advanced degrees from the Perimeter Institute and the Institute for Quantum Computing.
  • Trevor McCourt (Chief Technology Officer):
    McCourt leads Extropic’s technical strategy and engineering teams.
  • Christopher Chamberland (Principal Architect):
    Chamberland brings expertise in quantum and classical computing architectures.

Core Technologies

  • Thermodynamic Computing Chips:
    Extropic’s core innovation is a specialized chip that harnesses out-of-equilibrium thermodynamic systems—such as supercooled Josephson junctions—to perform probabilistic computations. These chips leverage the inherent noise and randomness of matter to accelerate energy-based models (EBMs) and generative AI workloads, dramatically reducing power consumption and enabling new forms of analog, stochastic, and adaptive computation.
  • Stochastic Analog Circuits:
    The chips feature parameterized stochastic analog circuits, which use physical noise as a computational resource, enabling efficient sampling and probabilistic inference for AI models.
  • AI Acceleration for LLMs:
    Extropic’s hardware is optimized for large language models and other generative AI systems, offering faster processing, scalability, and superior energy efficiency compared to conventional digital chips.
  • Edge Intelligence:
    The technology enables high-performance AI at the edge, supporting applications in national security, autonomous systems, and real-time decision-making with quantified uncertainty.

Key Capabilities

  • Thousand-fold improvements in energy efficiency and speed for AI and decision-making systems.
  • Native support for probabilistic and energy-based AI models, enabling superior pattern recognition and decision-making with less data and power.
  • Scalable integration into existing AI infrastructure, supporting both cloud and edge deployments.
  • Enhanced reliability, interpretability, and human-in-the-loop control for mission-critical AI applications.

Investors

  • Seed Funding: $14.1 million raised in March 2024.
  • Lead Investor: Kindred Ventures (Steve Jang).
  • Other Investors: Buckley Ventures, HOF Capital, Julian Capital, Marque VC, OSS Capital, Valor Equity Partners, Weekend Fund, and prominent angels including Aidan Gomez (Cohere), Amjad Masad (Replit), Arash Ferdowsi (Dropbox), Balaji Srinivasan, Garry Tan (YC), Naval Ravikant, Scott Belsky (Adobe), and Tobias Lütke (Shopify).
  • Company Status: Privately held; valuation not publicly disclosed.

Notable Clients

Specific commercial clients have not been publicly announced. Extropic’s technology is of strategic interest to sectors requiring advanced AI hardware, including national security, defense, and large-scale AI infrastructure providers. The company has been featured at major industry events and is actively collaborating with partners across the AI and computing ecosystem.

Competitors

Company NameDescription
GraphcoreUK-based startup developing Intelligence Processing Units (IPUs) for AI workloads, including LLMs.
Cerebras SystemsDeveloper of the Wafer Scale Engine, a chip for deep learning and AI acceleration.
NVIDIALeading provider of GPUs widely used for AI training and inference.
GoogleDeveloper of Tensor Processing Units (TPUs) and other AI hardware solutions.

More Resources

OODAloop Mentions and Resources