Odyssey, the AI lab founded by autonomous-driving veterans, has raised a $310 million Series B at a $1.45 billion valuation to accelerate its work on general world models - AI systems that learn to simulate and predict how the physical world behaves. The round, announced on June 17, 2026, was led by Natural Capital with participation from Amazon, AMD Ventures, GV, EQT and IQT.
A new class of foundation model
Unlike language models that predict the next word, a world model learns how objects move, how light falls and what happens after something is touched or pushed. Co-founder and CEO Oliver Cameron, who previously led product at Cruise and ran Voyage, said the field is approaching its "GPT-3 moment." Odyssey, founded in 2023 with Jeff Hawke, argues those capabilities could reshape robotics, science, healthcare, gaming and autonomous systems.
An Amazon-powered build-out
Alongside the raise, Odyssey named Amazon Web Services its preferred cloud provider and said it will train and deploy models on AWS Trainium chips, collaborating with Amazon's Annapurna Labs. Existing backers include Jeff Dean, Kyle Vogt, Qasar Younis, Garry Tan and Elad Gil, with Pixar co-founder Ed Catmull on the board. The capital, the company said, provides the compute and infrastructure to push the frontier of general world models.
The world-models race heats up
Odyssey's research stack already spans Odyssey-2 for physics-accurate simulation, the real-time multimodal Starchild-1, the multi-agent Agora-1 and the exploration framework PROWL. The round underscores surging investor appetite for systems that understand the physical world, days after Yann LeCun's AMI Labs raised a $1.03B seed for world models and amid releases such as NVIDIA's Cosmos 3. Such models are increasingly tied to embodied platforms like Genesis AI's Eno.
Reporting based on coverage from Odyssey, TechCrunch, Bloomberg and AI Insider.