Odyssey raised $310 million in a Series B at a $1.45 billion valuation, with Natural Capital leading and Amazon, AMD Ventures, and GV joining the round. The deal, announced June 17, 2026, pushes the world model startup's total funding to $337 million and gives it unicorn status less than three years after founding. As part of the deal, AWS becomes Odyssey's preferred cloud provider, and the company will optimize its models to run on Amazon's Trainium chips — a direct alternative to Nvidia's AI silicon.
Odyssey builds world models, a class of AI systems that simulate physical environments with accurate physics rather than generating text or chat responses. The startup is best known for producing rich, interactive video from text prompts, and it now offers models targeting use cases from video-game creation to robotics. The pitch to investors is that world models are the next frontier of generative AI — systems that don't just describe the world but render it in motion.
The founder résumés explain the bet. CEO Oliver Cameron co-founded autonomous vehicle startup Voyage, which was acquired by GM's Cruise, where he later served as VP of product. CTO Jeff Hawke was an engineer at UK self-driving company Wayve. Both spent years inside the data collection and simulation pipelines that self-driving cars depend on — the same primitives world models need at scale.
Key facts
- 01Odyssey raised a $310M Series B at a $1.45B valuation, led by Natural Capital with Amazon, AMD Ventures, and GV participating.
- 02The 2023-founded startup has now raised $337M in total funding.
- 03AWS is now Odyssey's preferred cloud provider, with models optimized to run on Trainium chips rather than Nvidia GPUs.
- 04Founders Oliver Cameron and Jeff Hawke come from self-driving — Cameron from Voyage and Cruise, Hawke from Wayve.
- 05Angel investors include Jeff Dean, Elad Gil, Garry Tan, Guillermo Rauch, and Cruise founder Kyle Vogt.
Odyssey's data approach borrows from Google Earth's playbook, but on foot. Rather than driving camera-equipped cars around, the startup sends people out with cameras strapped to their backs to capture real-world environments. That sidesteps some of the regulatory and access constraints that come with vehicle-mounted capture and lets the team collect interior spaces, sidewalks, and locations cars can't reach.
Founded in 2023, Odyssey has moved quickly from a research-oriented startup to a multi-product company. The Series B gives it the cash to scale training runs that world models demand — physics simulation at video frame rates is computationally heavier than next-token prediction, and the cost curve is steep.
The AWS Trainium commitment is the more strategic line in the announcement. Amazon has been pushing Trainium aggressively as a Nvidia alternative for training large models, and landing a high-profile world model customer gives the chip credibility outside Amazon's first-party workloads. For Odyssey, the deal likely comes with favorable compute pricing — a meaningful subsidy at this stage of scaling. AMD Ventures' participation hints that AMD's Instinct accelerators may also play a role over time.
The angel list reads like a who's-who of the AI and startup ecosystem: Google's Jeff Dean, investor Elad Gil, Y Combinator's Garry Tan, Vercel CEO Guillermo Rauch, and Cruise founder Kyle Vogt. Vogt's involvement is notable given Cameron's history at Cruise — the autonomous-vehicle network around Odyssey is dense, and that talent flow is a competitive moat in a field where simulation expertise is rare.
World models are crowded territory. Wayve, where Hawke previously worked, is building its own simulation stack. Nvidia, Google DeepMind, and a wave of robotics startups are all pursuing variants. The category is real, but no one has shipped a commercial breakout yet, and the gap between impressive demos and production-grade simulation is where most of these companies will be judged. Odyssey's bet on Trainium also carries technical risk — Nvidia's CUDA ecosystem remains the path of least resistance for most ML teams.
Odyssey's raise is a vote that world models are the next category investors want exposure to after the LLM gold rush. The Amazon-AMD alignment matters more than the headline valuation: it signals that hyperscalers see world simulation as the workload they need to win for the next compute cycle, and they're willing to subsidize promising startups onto non-Nvidia silicon to get there. If Odyssey delivers, it validates Trainium for frontier training. If it doesn't, the $1.45 billion mark will look early.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
