Odyssey AI — World models that generate interactive video in real time
Odyssey AI is an AI lab headquartered in Palo Alto that builds general-purpose world models. These models predict outcomes in physical space, unlike large language models that focus on text prediction. Odyssey's technology creates interactive video streams that respond to user inputs in real time, simulating real-world scenarios. Key applications include robotics training, gaming, autonomous vehicles, scientific research, and national security.
Get it by email
Companies like Odyssey AI go out every weekday morning. One-click unsubscribe.
- Round
- $310M Series B
- Investors
- Amazon, AMD Ventures, Eqt, Garry Tan, Guillermo Rauch, Gv, In-Q-Tel, Jeff Dean, Kyle Vogt, Natural Capital
- Sectors
- Artificial intelligence, Machine learning
- Logged
- June 19, 2026
- Website
- odyssey.ai
More in artificial intelligence and machine learning
$1B Growth · fal.ai
Fal — Inference for AI video, image and audio applications
fal is the leading platform for generative media, offering cutting-edge inference solutions for AI-powered video, image, and audio applications. The company partners with enterprises, research labs, and developers worldwide to accelerate the future of generative AI content creation.
Information technologySoftwareArtificial intelligence$100M+Machine learning · +2
$45M Series A · mountthor.com
Mount Thor — Cloud Macs and macOS VMs for AI agents and RL environments
Mount Thor runs cloud infrastructure built on Apple hardware: dedicated physical Macs and on-demand isolated macOS virtual machines for AI agents, RL environments and inference on Apple silicon. Founded in 2026 in San Francisco by Peter Voell, with a team that built infrastructure at Cerebras, CoreWeave, Fluidstack, Nebius, OpenAI and xAI, it deployed its first data center in under six months. The $45M is equity across a Seed (co-led by Chad Byers and Offline Ventures) and a Series A (co-led by CRV and Conviction), with CSC Leasing, Enea Capital, Long Journey Ventures and Upper90 participating.
Artificial intelligenceMachine learningCrvConvictionChad ByersOffline Ventures · +5 · +8
$25M Series A · luminal.com
Luminal — Compiler and inference cloud for AI models on GPUs and ASICs
Luminal builds a search-based compiler and inference cloud for AI models. The compiler tests mathematically equivalent ways to run a model on a given chip and picks the fastest, so workloads can run across GPUs and ASICs instead of depending on Nvidia alone. Founded by former Intel chip designer Joe Fioti with Matthew Gunton and Jake Stevens (Amazon, Apple), the San Francisco company says its cloud handles about 90 trillion tokens a month, four months after public launch. The $25M Series A was led by Standard Capital with Y Combinator, Felicis, SagaVC and Liquid, taking total funding to about $30.3M.
SoftwareArtificial intelligenceYCFelicisSaga VCLiquid · +2 · +5
AMD Ventures also backed River AI, Turing Enterprises, Hark