WillItSurvive
United States · United States

Odyssey

Startup building AI world models that simulate physical spaces for film, games, robotics.

Losing its protection
Assessed 29 July 2026 · Evidence as of 29 July 2026
The structural protection is eroding.
On the recordSince 29 July 2026 · Checkable from H2 2026 · Every call on the record Checkpoint: Google DeepMind moving Genie/Omni from research preview into full production inside Flow or YouTube Shorts
The call

Odyssey has money, talent, and a big idea, but no disclosed revenue, while Google DeepMind and better-funded rivals race toward the same world-model future with more reach and more cash.

Odyssey is a startup building artificial intelligence software that can generate and understand video. The company has raised a lot of money and hired smart people, but has not released a product customers pay for and does not have anything that would be hard for competitors to copy. Google owns a team called DeepMind that is already putting similar technology into YouTube and other apps that billions of people use every day, and other well-funded startups like World Labs and Runway are also racing to build the same thing. The biggest question is whether Google can turn its video AI into a product people actually use regularly in the next year or so, which would likely make it very hard for Odyssey to compete.

Why
01
AI pressure: Google DeepMind's Genie 3 / Omni world models. Genie already feeds real products like Flow and YouTube Shorts, giving Google distribution and scale data Odyssey does not have. Could AI do the core job? Yes, likely.
02
Bigger players: Google DeepMind. Google can fold world-model generation into existing video and creator products, turning Odyssey's core idea into a feature. Could a giant copy it? Yes, easily.
03
What protects it. Deep research talent, Ed Catmull on the board, and an AWS Trainium partnership give it credibility and cheaper compute, but no proven customers yet.
What would change this call
Google DeepMind's Genie/Omni exits research preview into Flow or YouTube Shorts at scale within 12 months. Would show DeepMind converting distribution advantage into a shipped rival product, squeezing Odyssey's window.
World Labs discloses Marble ARR or paying customer contracts. Would show a better-funded rival achieving commercial traction Odyssey has not yet shown.
Odyssey announces a forced Series C or discloses first enterprise customer and contract value. Would reveal whether Odyssey can convert capital into revenue before rivals do.
Key numbers
  • $310M Series B raised (2026-06-17)
  • $1.45B valuation after Series B (2026-06-17)
  • $2-4 per person-hour usage cost (2026-06-17)
Named risks
  • No disclosed revenue despite $1.45B valuation as of June 2026. Severity: high
  • Rivals hold 2 to 4 times more capital (World Labs, AMI Labs, Runway) in same capability race. Severity: high
  • Usage cost of $2 to $4 per person-hour creates ongoing capital-burn pressure. Severity: medium
  • Small team of ~46 to 55 employees competing against frontier-lab scale operations. Severity: medium
  • AWS Trainium exclusivity vs. Nvidia as investor creates chip-strategy tension. Severity: low
Rivals, as read in July 2026
RivalDirectionEvidence
Google DeepMindlosing toGenie/Omni shipping into Flow and YouTube Shorts with distribution scale as of June 2026
World Labsgaining onShipped commercial product Marble in Nov 2025, raised $1B+ at ~$5B valuation
AMI Labsgaining onRaised $1.03B seed at $3.5B valuation, largest European seed on record, May 2026
RunwaycontestingRaised $315M at $5.3B valuation, launched GWM-1 universal world model, Dec 2025
Questions people ask
Is Odyssey profitable in 2026?
No revenue has been disclosed by Odyssey in any funding coverage as of its June 17, 2026 Series B announcement. The Financial Times reported usage pricing of $2 to $4 per person-hour on Nvidia H200/B200 chips, but that is a cost signal, not confirmed revenue or profit.
Could Google DeepMind replace Odyssey's world models?
Google DeepMind's Genie and reported Omni model are shipping into Flow and YouTube Shorts as of June 2026, giving DeepMind distribution and Street View scale data that Odyssey lacks. Odyssey's product is itself a frontier model competing directly on capability, not a protected layer.
How much has Odyssey raised and what is it worth?
Odyssey closed a $310 million Series B on June 17, 2026, led by Natural Capital with Amazon, AMD Ventures, GV, EQT and In-Q-Tel, valuing the company at $1.45 billion post-money and bringing total funding to $337 million.
Who is Odyssey's biggest competitor?
World Labs, led by Fei-Fei Li, shipped its commercial product Marble in November 2025 and has raised over $1 billion at valuations discussed up to $5 billion, roughly 3.5x Odyssey's $1.45 billion valuation as of June 2026.
What products has Odyssey actually shipped?
Odyssey launched Explorer in December 2024 (image-to-3D-world generation), followed by Odyssey-2/Odyssey-2 Max (real-time simulation at ~50ms latency), Starchild-1, and Agora-1, demoed as a live multiplayer deathmatch at RAAIS London on June 12, 2026.
Does Odyssey have any customers?
London virtual-production house Garden Studios tested Explorer-generated worlds for films, commercials and music videos, reported in April 2025 (originally December 2024). No customer counts, ARR, or revenue figures have been disclosed as of June 2026.
Timeline
  • 2026-06-17Odyssey closes $310M Series B at $1.45B valuation, announces AWS Trainium partnership
  • 2026-06-12Agora-1 multi-agent world model demoed as live multiplayer deathmatch at RAAIS London
  • 2025-11-12Rival World Labs ships commercial world-model product Marble
  • 2024-12-18Odyssey launches Explorer image-to-3D-world model; Ed Catmull joins board
  • 2024-12Odyssey's total funding reaches $27M via seed extensions
  • 2024-07-08Odyssey raises $9M seed round led by GV, with DCVC and Air Street Capital
Revision history
29 July 2026
First assessed. Losing its protection. The text above is fixed as of this date and is not edited afterwards.
In the same band
Sources