Odyssey Reaches $1.45 Billion Valuation in Amazon-Backed Funding Round
Odyssey is building AI world models designed to simulate physical environments and physics, moving beyond text-based generation. The startup, which recently raised a $310 million Series B, is optimizing its infrastructure for AWS Trainium chips.
Odyssey is building what it calls world models—AI systems designed to gather data from the physical world and simulate it with accurate physics. According to TechCrunch, the Menlo Park-based company, founded in 2023, is positioning its technology as a successor to text- and chat-based large language models. The startup creates interactive video from text prompts and targets a range of applications, including robotics and video game development. By simulating real-world environments, Odyssey aims to provide a foundation for spatial AI, distinguishing its output from the standard text-prediction capabilities of current generative models.
The company’s data acquisition strategy mirrors early mapping efforts. According to TechCrunch, Odyssey gathers its training data by sending people out with cameras strapped to their backs, a physical approach reminiscent of how Google Earth collected its initial imagery. This focus on spatial data is rooted in the founders' backgrounds: CEO Oliver Cameron previously co-founded the autonomous vehicle startup Voyage, later acquired by Cruise, while CTO Jeff Hawke worked as an engineer at the self-driving startup Wayve. This operational history informs the company’s current focus on building models that understand and interact with physical environments.
As the startup scales its simulation engine, it is consolidating its infrastructure strategy. According to TechCrunch, Odyssey has selected AWS as its preferred cloud provider. The partnership involves optimizing its world models to run on AWS’s Trainium chips, which are designed to compete with Nvidia’s hardware. This alignment suggests a strategic dependency on Amazon’s compute stack to handle the heavy processing requirements inherent in physical-world simulation. While the company has secured a $310 million Series B round led by Natural Capital, the shift to AWS hardware highlights the capital-intensive nature of training models that require high-fidelity physics and spatial data.
The $310 million Series B round, which Crunchbase News reports brings the company’s total funding to $337 million, places Odyssey at a $1.45 billion valuation. The round saw participation from a broad group of investors, including AMD Ventures, GV, EQT, IQT, and SignalRank, alongside a list of angel investors such as Jeff Dean, Elad Gil, and Cruise founder Kyle Vogt. This funding arrives as the broader AI market seeks alternatives to the commoditized large language model space. By focusing on multimodal simulations of real-world environments, Odyssey is attempting to carve out a niche in AI infrastructure that extends beyond simple text generation.
Despite the influx of capital, the startup faces the challenge of proving the utility of its world models across diverse industries. While the company claims its technology applies to use cases ranging from video game creation to robotics, it must now demonstrate that its simulation engine can provide tangible value in these sectors. The reliance on specialized hardware like Trainium chips and the high cost of collecting physical-world data suggest that Odyssey’s path to commercial viability depends on its ability to scale its simulation platform efficiently. The company’s next phase will likely be defined by how effectively it translates its research into deployable enterprise software.