Odyssey hits a $1.45 billion valuation with Amazon backing

Published: 2026-06-19T08:25:57.000Z

For the past few years the AI industry has been almost entirely consumed with language, the models that power chatbots. Odyssey is betting the next frontier lies elsewhere, in teaching machines to understand the physical world. The AI lab has raised $310 million at a $1.45 billion valuation, minting it as a unicorn.

For the past few years the AI industry has been almost entirely consumed with language, the models that predict the next word and power chatbots like ChatGPT and Gemini. Odyssey is betting the next frontier lies somewhere else entirely, in teaching machines to understand the physical world rather than just text. The AI lab has raised $310 million in a Series B round that values it at $1.45 billion, officially minting it as a unicorn. The round was led by Natural Capital, with a notably heavyweight cast of backers including Amazon, AMD Ventures, Google Ventures, EQT and the CIA linked fund In-Q-Tel, alongside individuals like Google chief scientist Jeff Dean and Y Combinator chief Garry Tan. What Odyssey is building is a category researchers call world models, and the distinction from today's dominant systems is the whole point. Rather than predicting the next word in a sentence, Odyssey's models are trained to predict what happens next in the physical world, learning concepts like movement, space, object interactions and cause and effect. The aim is AI that can simulate and anticipate real environments the way a language model anticipates text. Co-founder and chief executive Oliver Cameron frames world models as a new class of foundation model, one that can understand and simulate the world itself, and he has openly described the goal as reaching a GPT-3 moment for the field, the inflection point where the technology suddenly becomes broadly useful. The founders' background explains why this bet is credible rather than speculative. Odyssey was started in 2023 by Cameron and Jeff Hawke, both veterans of the autonomous driving world, a field that has spent more than a decade trying to teach machines to perceive and predict the messy physical environment from a moving vehicle. Cameron previously co-founded the self driving startup Voyage, which was acquired by GM's Cruise. That heritage shows in how Odyssey gathers training data, reportedly deploying people with body mounted cameras to capture real world environments at scale, an approach reminiscent of how Google Earth was built. The company already offers models spanning video game creation, robotics and interactive video generated from text prompts, bringing its total funding to $337 million. The infrastructure side of the deal is just as telling. Alongside the raise, Odyssey signed a strategic agreement making Amazon Web Services its preferred cloud provider, with access to AWS Trainium chips, Amazon's custom AI accelerators built for large scale machine learning. World models are enormously compute hungry, since simulating physics in real time is far heavier than generating text, so locking in cloud capacity and a chip partner is as strategically important as the cash itself. The regional read points to real relevance for the Gulf. Across the Middle East and North Africa, governments are investing heavily in robotics, autonomous mobility and physical AI, with Saudi Arabia and the UAE funding automation for warehouses, ports, construction and mega projects like NEOM. World models are precisely the kind of foundational technology those ambitions depend on, since robots and autonomous systems need to understand and predict physical environments to work safely. As the region builds the infrastructure of physical AI, labs like Odyssey are developing the brains that could eventually run inside it.