# Pasteur Labs & ISI

> Pasteur Labs & ISI is a New York-based public-benefit deep-tech company building 'Simulation Intelligence' - the systematic merger of AI, physics simulation, and scientific computing. Its AI-native, GPU-accelerated, end-to-end differentiable physics platform gives engineers and scientists an in-silico playground to run thousands of what-if scenarios and shrink the gap between digital models and physical reality. Founded in 2021 by AI researcher Alexander Lavin and staffed by veterans of DeepMind, NASA, CERN, Tesla, Nvidia and Ansys, the company targets industrial R&D, energy, climate, and space & defense, and stewards the open-source Tesseract framework for differentiable scientific computing.

- **Founded:** 2021
- **Headquarters:** New York, New York, United States
- **Founders:** Alexander Lavin (Founder & CEO)
- **Team size:** ~43 employees (reported; roughly 25+ full-time engineers, scientists and leaders)
- **Products:** Simulation Intelligence (SI) Platform, Tesseract, Differentiable physics simulators, Applied R&D services
- **Notable:** Authored the foundational 'Simulation Intelligence' framework paper (Lavin et al., 2021), Co-authored 'Technology Readiness Levels for Machine Learning Systems' (MLTRL), published in Nature Communications (2022), Raised a $10.5M seed round (2023); ~$12.8M total funding reported

## Products & services

- **Simulation Intelligence (SI) Platform** — An AI-native, GPU-accelerated in-silico playground for human-machine teams. Uses patent-pending end-to-end differentiable physics programs and CAxML data interfaces so computational engineers, scientists and ML/AI specialists can co-design and run thousands of scenarios at speed.
- **Tesseract** — Free, open-source (Apache 2.0) framework for differentiable scientific computing. Packages scientific software into self-contained, portable components that run on laptops, cloud, or HPC with identical results, exposing CLI, REST API and Python SDK interfaces plus derivatives for end-to-end optimization.
- **Differentiable physics simulators** — Data-driven, GPU-native, automatically differentiable multiphysics simulators that enable gradient-based design, surrogate-based optimization and uncertainty-aware computation.
- **Applied R&D services** — Use-inspired research and simulation testbeds for industrial R&D, energy security, nuclear/inertial fusion modeling, and space & defense autonomous systems.

## Achievements

- Authored the foundational 'Simulation Intelligence' framework paper (Lavin et al., 2021)
- Co-authored 'Technology Readiness Levels for Machine Learning Systems' (MLTRL), published in Nature Communications (2022)
- Raised a $10.5M seed round (2023); ~$12.8M total funding reported
- Released Tesseract Core as open-source (Apache 2.0) in 2025 and ran the Tesseract Hackathon 2025
- Acquired defense-tech startup FOSAI in August 2025
- Structured as a public-benefit corporation committing 50% of resources to its public mission

## Latest updates

- **2025-08** — Acquired AI startup FOSAI to accelerate AI-driven physics simulation for space & defense applications.
- **2025-03** — Released Tesseract Core as free, open-source software (Apache 2.0) for differentiable scientific computing.
- **2025-01** — Announced late-2025 public launch of the Simulation Intelligence Platform; pre-release versions deployed to dual-use customers.
- **2023-09** — Closed a $10.5M seed round.

## Links

- Website: https://simulation.science
- LinkedIn: https://www.linkedin.com/company/pasteur-isi
- GitHub: https://github.com/pasteurlabs

---

Profile page: https://yespress.io/pasteur-labs-isi
Published by YesPress — https://yespress.io
Last updated: 2026-07-03