Startup Odyssey raised $310 million at a $1.45 billion valuation, Amazon among investors

Odyssey, an AI-powered world models startup, raised $310 million in a Series B round and achieved a $1.45 billion valuation. The deal includes Amazon, AMD Ventures, GV and other investors, with the company rapidly becoming one of the most notable players in the AI market beyond traditional LLMs.

Funding Round and Valuation

On June 17, 2026, the company announced a new round led by Natural Capital. For Odyssey, this is not just a significant capital injection, but a transition into the unicorn club: the startup's valuation reached $1.45 billion.

Against the backdrop of a general cooling in parts of the AI market, such a round looks like a strong signal that investors are willing to bet not only on chatbots and assistants, but also on more complex models that attempt to describe the physical world. Along with venture funds, the company attracted a notable group of angel investors. Among them are Jeff Dean, Elad Gil, Harry Tran, Guillermo Rauch, and Cruise founder Kyle Vogt.

Following this round, the total capital raised by Odyssey reached $337 million. For a startup founded only in 2023, this is very rapid capital accumulation, especially considering it is an expensive field requiring significant computational power and proprietary data.

• $310 million — size of the new Series B round
• $1.45 billion — Odyssey's current valuation
• $337 million — total amount of capital raised
• Natural Capital — round leader
• Amazon, AMD Ventures, and GV — among the deal participants

Why World Models Matter

Odyssey operates in the category of world models. Unlike large language models, which primarily rely on text and dialogue, such systems learn from physical world data and attempt to reproduce environments while accounting for the laws of physics. This is why there is such interest around the direction: if a model truly understands space, movement, objects, and causal relationships, it opens up applications poorly covered by conventional LLMs.

According to the company, its models can be used in various scenarios — from video game creation to robotics. However, the broader audience likely knows Odyssey primarily as the developer of technology that generates rich, interactive video based on text prompts. This is no longer just a video that plays along a pre-prepared script, but an attempt to construct an environment that can be interacted with more fluidly and flexibly.

The startup's approach to data collection is also instructive. Odyssey essentially replicates the logic that Google Earth once used: the company sends people equipped with cameras mounted on their backs to gather material from the real world. This method aligns well with the founders' background.

Chief Executive Officer Oliver Cameron was previously co-founder and CEO of autonomous vehicle startup Voyage, which was later acquired by Cruise, where he later became Vice President of Product. Chief Technology Officer Jeff Hawke worked as an engineer at British autonomous vehicle startup Wayve.

Betting on AWS

Amazon's participation in this round is important not only as a name on the investor list. Odyssey announced that AWS will become its priority cloud provider, and the models will be optimized for Trainium chips. For Amazon, this is a chance to demonstrate that its AI infrastructure can be appealing not only to corporate clients but also to ambitious labs building new types of models and seeking alternatives to the Nvidia ecosystem.

For Odyssey itself, such a partnership also looks strategic. A company working with physical world models needs expensive computing, a stable cloud stack, and a clear scaling partner. If optimization for Trainium proves successful, the startup gains not just financing but also stronger infrastructure support.

In today's market, this is almost as important as the size of the round itself: AI companies win not only through models, but also through what they use to train and run those models on.

What This Means

Large rounds around Odyssey show that the market is beginning to view world models as a distinct and promising branch of AI, rather than a side experiment alongside LLMs. If investor bets pay off, the next wave of AI competition will not only be about the best text interface, but also about systems that can plausibly model space, movement, and object behavior in the real world.