World-Model Startup Odyssey Raises $310 Million at $1.45 Billion Valuation, Picks AWS Trainium Months After Taking Nvidia's Money
Odyssey closed a $310 million Series B led by Natural Capital with backing from Amazon, GV, and AMD Ventures, naming AWS its preferred cloud after an Nvidia-backed Series A in February.
Editor's Note ·
- Clarification:
- Three of the article's four sources — odyssey.ml (the company's own announcement), techfundingnews.com, and unite.ai — are not on The Machine Herald's vetted source allowlist. The Chief Editor verified the cited claims against the captured snapshots of odyssey.ml and techfundingnews.com; the funding figures and all quotes attributed to these outlets were confirmed verbatim. Readers should note that odyssey.ml is a primary-party (company) source for this funding announcement.
- Clarification:
- The unite.ai source, cited for Odyssey's model lineup (Odyssey-2, Odyssey-2 Max, Starchild-1, Agora-1, PROWL), blocked automated archiving (HTTP 403), so no snapshot of it exists in the provenance record. That same model lineup is independently confirmed by Odyssey's own announcement (odyssey.ml), which was captured and verified, so no claim in the article is left unverified.
Overview
Odyssey, an AI research company building so-called world models, has raised a $310 million Series B at a $1.45 billion valuation, according to the company’s announcement. The round was led by Natural Capital, with participation from Amazon, GV, AMD Ventures, EQT, and IQT, the company said. TechCrunch reported that the new round brings Odyssey’s total raised to date to $337 million.
The financing arrives with a notable absence. Tech Funding News reported that “NVentures, Nvidia’s venture capital arm, backed Odyssey’s Series A in February 2026,” but that for the new round “NVIDIA is not part of the Series B group. Instead, AMD Ventures is a new shareholder, and AWS Trainium is now the chip of choice.”
What We Know
Odyssey was founded in 2023 and is led by CEO Oliver Cameron, formerly co-founder and CEO of the self-driving startup Voyage and later a VP of product at Cruise, and CTO Jeff Hawke, a former engineer at Wayve, according to TechCrunch. TechCrunch described the company as “perhaps best known for producing rich, interactive video from text prompts,” generating training data in part by sending “people out with cameras strapped to their backs.”
As part of the deal, Odyssey named Amazon Web Services its preferred cloud provider and said it would optimize its models to run on AWS Trainium chips, according to the company. Unite.AI reported that the company’s research and product efforts include Odyssey-2 and Odyssey-2 Max, a model focused on physics-accurate world simulation; Starchild-1, described as a real-time multimodal world model; Agora-1, aimed at multi-agent interaction; and PROWL, a reinforcement learning framework.
In the company’s announcement, Ron Diamant of Amazon said, “World models represent one of the most demanding workloads in AI—they require massive compute throughput with tight latency constraints.” Jay Zaveri of Natural Capital, the lead investor, said of world models, “We believe they have the potential to help define AI beyond language models.” Luna Schmid of GV said, “World models are now a multi-billion-dollar category, and Odyssey has been leading the way,” per the announcement.
What We Don’t Know
Neither Odyssey nor its investors detailed why Nvidia’s venture arm did not return for the Series B after backing the February round. Tech Funding News noted the uncertainty directly: “It is unclear whether this change is due to a real belief in Amazon’s technology or simply better deal terms in a highly competitive market.” The company has not published benchmark results or commercial deployment figures that would allow a direct comparison of its models against rival world-model efforts.
Analysis
World models — systems that aim to understand and simulate physical reality rather than predict the next token of text — have drawn growing investor interest as a category distinct from large language models. According to Odyssey, the company frames the applications broadly, spanning robotics, science, healthcare, education, gaming, and defense. The chip decision is the most concrete signal in the round: by committing to AWS Trainium, an Nvidia-backed startup is steering a flagship workload toward Amazon’s in-house silicon, a choice Tech Funding News characterized as “Amazon’s answer to Nvidia’s dominance in AI computing.”