Odyssey raises $310M Series B; pivots from NVIDIA to AWS Trainium for world models
Odyssey, a Palo Alto AI lab building physics-aware world models, closed a $310 million Series B at a $1.45 billion valuation on June 17. The round was led by Natural Capital with participation from Amazon, AMD Ventures, Google Ventures, EQT, In-Q-Tel, and angels including Google chief scientist Jeff Dean, Y Combinator's Garry Tan, and Cruise founder Kyle Vogt. The startup has now raised $337 million total.
The headline change: AWS becomes Odyssey's preferred cloud provider, and the company will optimize its models for Amazon's Trainium AI chips. Just four months earlier, Nvidia's venture arm (NVentures) backed Odyssey's Series A—Nvidia is absent from this round. The pivot underscores mounting competition in AI infrastructure, with Amazon and AMD offering alternative silicon for the high-throughput, low-latency workloads that real-time world simulation demands.
Odyssey was founded in late 2023 by Oliver Cameron (former VP of Product at Cruise) and Jeff Hawke (founding engineer at Wayve). The team develops general-purpose world models—causal, multimodal AI systems that simulate persistent physical environments where multiple agents can act simultaneously. Recent milestones include Odyssey-2 Max (physics-accurate simulation), Starchild-1 (first real-time multimodal world model), Agora-1 (multi-agent interaction), and PROWL (active exploration learning).
For practitioners: This marks a critical moment in the world-models arms race. Odyssey is betting $310M and its chip roadmap on breaking Nvidia's stranglehold on frontier AI compute. With 55 employees and high per-head capital intensity, Odyssey is pursuing robotics, gaming, and science—and the defense mentions (In-Q-Tel's participation, applications in autonomous systems) flag this as a dual-use technology to watch for national security deployment timelines.