Odyssey, a Los Angeles AI startup building 'world models,' raised a $310 million Series B at a $1.45 billion valuation, vaulting it to unicorn status and bringing total funding to roughly $337 million. The round was led by Natural Capital, with a deep bench of strategic and angel backers including Amazon, AMD Ventures, GV, Jeff Dean, Elad Gil, Garry Tan, Guillermo Rauch, and Kyle Vogt. The company was founded in 2023 by Oliver Cameron and Jeff Hawke, both veterans of the self-driving industry.
World models are the heart of the bet. Rather than predicting the next token of text, Odyssey's systems learn to simulate physical environments -- generating interactive video and scenes that obey real-world physics. That capability is foundational for the categories investors are most excited about heading into the back half of 2026: robotics, autonomous systems, and immersive interactive media, all of which need a model of the world rather than a model of language.
“The company was founded in 2023 by Oliver Cameron and Jeff Hawke, both veterans of the self-driving industry.”
The cap table is the strategic tell. By bringing in Amazon and AMD Ventures and naming AWS its preferred cloud -- optimizing its models to run on AWS Trainium silicon -- Odyssey is aligning itself with the anti-Nvidia compute coalition, a notable pivot for a company that had previously taken Nvidia's capital. It's a reminder that the question of whose chips and whose cloud train the next generation of models is now a competitive battleground, not a foregone conclusion.
The round lands at the top of an otherwise slower week for large deals, leading a lineup that skewed heavily toward AI infrastructure, defense, and biotech. Odyssey's scale -- the single biggest round of the week by a wide margin -- underscores how decisively capital is rotating toward physical AI as the pure-language-model trade gets crowded and commoditized.
For founders, the signal is clear: the frontier is moving off the screen. The companies that can give machines a usable, physically grounded simulation of reality are positioned to power the robotics and autonomy wave, and investors are willing to underwrite that vision at nine figures before the revenue fully materializes. The risk, as always with frontier bets, is timeline -- world models are early, and the gap between a stunning demo and a deployable product is exactly where these valuations get tested.