Amazon, AMD Back AI Startup Odyssey in $310M Series B

What happened

Odyssey, an AI lab developing world models, raised $310 million in a Series B round at a $1.45 billion post-money valuation. Natural Capital led the round, with Amazon, AMD Ventures, GV, EQT, and In-Q-Tel joining as investors. The company has now raised $337 million to date. Alongside the funding, Odyssey announced a strategic partnership with Amazon Web Services (AWS), designating AWS as its preferred cloud provider. The partnership gives Odyssey access to AWS's Trainium chips, custom silicon designed for training and deploying AI models at scale. Odyssey will optimize its models to run on Trainium and collaborate with AWS on joint research and commercialization efforts. The startup, founded in 2023 by CEO Oliver Cameron and CTO Jeff Hawke-both veterans of the autonomous vehicle industry-has released several research systems over the past three years. These include Odyssey-2 Max, which advanced physics accuracy in world simulation; Starchild-1, the first real-time multimodal world model; and Agora-1, which introduced multi-agent interactivity in shared simulated environments. World models differ from text-based large language models by building internal representations of reality, including physics, causality, and time, enabling them to simulate or predict outcomes before taking action.

Why it matters

For AMD, this investment positions the company in an emerging segment of AI beyond chatbots and language models. World models represent compute-intensive workloads that require specialized hardware for training and inference. AMD's participation through AMD Ventures aligns with its broader strategy to compete in AI accelerator markets, where it faces rivals like Nvidia. While Odyssey's partnership focuses on AWS's Trainium chips, the startup's growth could create future opportunities for AMD's AI hardware offerings, particularly in robotics, autonomous systems, and gaming-sectors where AMD already has established customer bases. The investment also signals AMD's recognition that simulation-based AI is becoming a critical frontier, with applications extending from physical AI agents to virtual environments. Odyssey's approach-gathering real-world data with camera-equipped personnel, similar to Google Earth's methodology-underscores the infrastructure and compute demands of world models. For AMD, exposure to this category could inform hardware development priorities as demand for physics-accurate simulation grows across industries.

Bigger picture

World models have emerged as a major category in AI investment, with multiple startups raising large rounds in recent months. Decart.ai raised $300 million at nearly a $4 billion valuation, AMI Labs (backed by AI pioneer Yann LeCun) raised $1.03 billion in March, and Runway AI raised $315 million in February. Google has also entered the space with Project Genie, a tool combining image generation and world models to create three-dimensional environments from prompts. The competitive landscape reflects growing investor conviction that simulation-based AI is the next frontier beyond large language models. These systems are essential for training physical AI agents, robotics, and autonomous vehicles, where simulating environments improves safety, reduces training time, and prepares systems for unexpected conditions. Odyssey's partnership with AWS highlights the importance of specialized chips for world model workloads, which require massive compute throughput with tight latency constraints. The involvement of In-Q-Tel, the venture arm of the US intelligence community, suggests potential government and defense applications for simulation technologies. For chip makers like AMD and Nvidia, world models represent a new class of customer demanding purpose-built AI accelerators capable of handling real-time physics simulation at scale.

What to watch

Monitor whether Odyssey expands its hardware partnerships beyond AWS, potentially creating opportunities for AMD's AI accelerators in production workloads. Watch for commercial deployments of Odyssey's models in robotics, gaming, or autonomous systems, which could validate demand for world model infrastructure. Track competitive developments from Google, Decart.ai, AMI Labs, and Runway AI, as well as broader adoption of simulation-based AI across industries. Pay attention to AMD's broader AI strategy and whether investments like this translate into hardware design wins or ecosystem partnerships. Finally, observe how AWS's Trainium chip performs in world model benchmarks, as this could influence future infrastructure decisions for simulation-heavy AI workloads.