Odyssey, the AI startup building real-time world models, has raised $310 million in a Series B round that brings Amazon into one of the field’s most closely watched bets.

The company said the financing values it at $1.45 billion. Natural Capital led the round, while Amazon, AMD Ventures, Google’s GV, EQT and In-Q-Tel also participated. As part of the deal, AWS will serve as Odyssey’s preferred cloud provider and supply its Trainium chips.

The participation of Amazon and AMD stands out because Nvidia, which backed Odyssey’s earlier funding round, is absent this time. Four months ago, Nvidia’s venture arm, NVentures, joined Odyssey’s Series A. The latest round instead aligns Odyssey with companies that are trying to offer alternatives to Nvidia’s dominance in AI infrastructure.

Odyssey’s chief executive and co-founder, Oliver Cameron, said the financing gives the company the compute, infrastructure and partnerships needed to push forward general world models. He framed the company’s goal as reaching a breakthrough comparable to the early scaling moment that helped define large language models.

World models aim to simulate how physical environments evolve over time. Rather than focusing on text, these systems are designed to understand cause and effect, object permanence and movement in 3D spaces. In practice, that could let a robotics company test thousands of scenarios in simulation before deploying systems into the real world.

The company was founded in late 2023 by Cameron, who previously worked at Cruise and Voyage, and Jeff Hawke, who was an early engineer at Wayve. Odyssey says its team of about 55 people is spread across Palo Alto, London and Zurich, and includes veterans of DeepMind, Tesla, Waymo and Apple.

Odyssey has been developing products such as Odyssey-2 Max, which is focused on physics simulation, and Agora-1, a system that allows multiple agents to act within a shared environment. The company has also pointed to potential applications beyond entertainment, including robotics and defense.

The latest funding round arrives as interest in world models intensifies across the AI sector. Runway recently reached a $5.3 billion valuation after a $315 million raise, while World Labs, founded by Fei-Fei Li, secured $230 million. Google DeepMind’s Genie has already found use at Waymo, and Yann LeCun’s new lab is reportedly seeking funding at a multibillion-euro valuation before releasing a product.

Odyssey’s strategy is especially ambitious given its size and stage. The startup has raised only about $27 million before this round, making the jump to a $1.45 billion valuation a sign of investor confidence in the field even though commercial returns are still unproven.

The mix of backers also hints at broader competition in AI hardware and cloud infrastructure. Amazon’s Trainium chips are designed for high-throughput AI workloads, and the company’s cloud business has been pitching them as an alternative to Nvidia-powered systems. For Odyssey, the arrangement provides access to computing capacity needed for simulation-heavy development, while for Amazon it adds another startup showcase for its in-house silicon.

With In-Q-Tel also participating, Odyssey’s focus on simulation may extend into government and security use cases, alongside robotics and gaming. The round suggests that, as world models draw more capital, the race is no longer only about who can build them, but also who can supply the infrastructure that keeps them running.