Odyssey, an AI lab founded by self-driving car veterans, announced on June 17, 2026 that it has raised a $310 million Series B at a $1.45 billion valuation. The funding will accelerate the company's work on general world models — AI systems that learn to simulate and interact with the physical world rather than merely generate text or images.
Key Highlights
- $310 million Series B at a $1.45 billion valuation, bringing total funding to roughly $337 million
- Natural Capital led the round, with Amazon, AMD Ventures, GV, EQT, and IQT participating
- Backed by industry leaders including Google chief scientist Jeff Dean, Y Combinator CEO Garry Tan, Vercel CEO Guillermo Rauch, and Cruise founder Kyle Vogt
- AWS becomes preferred cloud provider, with models optimized for Amazon's Trainium chips
- Founded in 2023 by autonomous-vehicle veterans Oliver Cameron and Jeff Hawke
Details
World models are a distinct branch of AI. Where large language models predict the next token of text, a world model learns the underlying dynamics of an environment — physics, cause and effect, how a scene evolves over time — and can generate or simulate that environment interactively. Odyssey describes its mission as building "causal, multimodal systems that learn to predict and interact with the world over long horizons."
The company unveiled a family of models alongside the raise. Odyssey-2 Max is a general-purpose world model with advanced physics accuracy. Starchild-1 is described as the first real-time multimodal world model, combining multiple sensory inputs. Agora-1 is a multi-agent world model that enables shared, real-time simulations. PROWL is a reinforcement-learning framework that uses active exploration to improve the models.
To gather the training data needed to simulate real environments accurately, Odyssey has reportedly sent individuals on foot with cameras strapped to their backs to map physical spaces — a labor-intensive approach that underscores how capital-hungry world simulation can be.
Impact
The raise signals that world models have matured into a serious investment category. "World models are now a multi-billion-dollar category, and Odyssey has been leading the way," said Luna Schmid, a partner at GV.
The deal also reflects the industry-wide scramble to reduce dependence on Nvidia hardware. By naming AWS its preferred cloud provider and committing to optimize models on Amazon's Annapurna Labs Trainium chips, Odyssey joins a growing list of AI labs diversifying their compute supply.
Background
Odyssey is part of a wave of well-funded world-model labs. The space drew major attention earlier in 2026 when Turing Award winner Yann LeCun launched AMI Labs with a record seed round to pursue a similar goal through a different architecture. Odyssey's founders, Oliver Cameron and Jeff Hawke, previously worked on autonomous-driving systems, where simulating the physical world is a core engineering problem.
What's Next
Odyssey says the new capital will fund continued research and the path to commercial deployment across robotics, science, healthcare, education, gaming, and defense. With AWS infrastructure and Trainium optimization in place, the company is positioning its world models as a foundational layer that other applications can build on.
Source: Odyssey