Odyssey Brief
B2B SaaS·World Models·Series B·🇺🇸🇬🇧🇨🇭
Summary
Odyssey is an AI lab building general-purpose, multimodal, causal world models aimed at long-horizon simulation and control (robotics/research use-cases), differentiated by productization and API distribution versus research-first rivals. The company closed a $310M Series B on June 17, 2026 (total disclosed funding $346M, $1.45B post-money) and announced a preferred-cloud partnership with AWS, signaling significant strategic backing from investors that include Amazon, AMD Ventures, EQT, IQT and Jeff Dean. Shipping Odyssey-2 Pro gives it a commercial lead over research labs such as AMI Labs and positions it against domain-specific simulators like Decart and World Labs, trading vertical fidelity for generality. The clearest tension: strong capital and infra partnerships set high growth expectations, but public commercial signals are thin (one published Samsung Next testimonial, no ARR or customer counts and only usage-based pricing disclosed), making go-to-market execution the single biggest determinant of whether product differentiation converts to scale.
Bull case
- Closed a $310M Series B (June 17, 2026) led by Natural Capital (total disclosed funding $346M) and a $1.45B post-money valuation — signals deep investor conviction and substantial runway for scaling product and GTM.
- Strategic infrastructure and investor endorsements: a preferred-cloud partnership with AWS plus investors including Amazon, AMD Ventures, EQT, IQT and Jeff Dean provide both cloud access and high-signal validation for enterprise customers and partners.
- Product + commercial lead vs research labs: Odyssey already ships Odyssey-2 Pro, giving a go-to-market advantage over research-first competitors like AMI Labs and enabling API-based distribution for robotics and long-horizon simulation use cases.
- Compact, senior team and global footprint: ~55 employees with offices in Palo Alto, London and Zurich and named leadership (Oliver Cameron, Jeff Hawke, James Grieve, Jessica Inman) consistent with a focused growth-stage lab.
Watch-outs
- Commercial traction is opaque: only one published customer testimonial (Andy Duong, Samsung Next), no disclosed ARR, ACV, or customer counts, and pricing is stated as usage_based with no unit economics — hard to validate demand or LTV/CAC dynamics.
- Crowded, bifurcated competitive landscape: domain-specific simulators (Decart for automotive, World Labs/Marble for persistent 3D worlds, Genie 3, World Simulator AI) may win on vertical fidelity or incumbent integrations, complicating Odyssey’s generalist go-to-market.
- Limited spontaneous visibility: the company is not routinely surfaced in unprompted category/recommendation outputs and public search returns noise from unrelated entities named “Odyssey,” suggesting weaker organic mindshare despite funding.
- Execution pressure from expectations: the large Series B and $1.45B valuation create high growth/monetization expectations; without published commercial metrics, scaling sales and proving unit economics is the primary execution risk.
From their website
Stats they publish
Pricing
Not publicly disclosed. The company website does not list its pricing on any of the 6 pages we read.
Product suite
- Odyssey-2
Company's most advanced general-purpose, multimodal world model for physically accurate long-horizon simulation.
- Starchild-1
Multimodal world model that learns from richer, non-visual inputs.
- Agora-1
Multi-agent world model enabling real-time interactions among human and AI participants in the same simulated environment.
- PROWL
Reinforcement-learning adversarial framework where an RL agent explores environments to improve world-model performance.
Target segments
Testimonials
“We were impressed with the rapid technical advances demonstrated by Odyssey-2 Pro, showing promising progress towards interactive world simulation, and the early signs of teaching artificial intelligence true cause-and-effect.”
- Trade name
- Odyssey Systems, Inc.
- Registered address
- Odyssey Systems, Inc. 1300 El Camino Real, Suite 100, Menlo Park, CA 94025
- [email protected]
Crunchbase
verified profile
- The Week’s 10 Biggest Funding Rounds: World-Model Startup Odyssey Leads With $310M In Slower Week For Large DealsCrunchbase News · 2026-06-18
- Odyssey Raises $310M in Series B Funding at $1.45 Billion ValuationFinSMEs · 2026-06-18
- World-model AI startup Odyssey secures $310Mn in funding at a $1.45Bn valuationThe Tech Portal · 2026-06-18
- AI world model startup Odyssey lands $310m in Series BVerdict · 2026-06-18
- Odyssey, 1.45 milyar dolar değerleme üzerinden 310 milyon dolar yatırım aldıWebrazzi · 2026-06-18
- Valued at 9.5 billion, Jensen Huang invested in a world model startup36kr.com · 2026-06-18
- AI lab Odyssey valued at $1.45 billion in latest funding roundMarketScreener · 2026-06-17
- Amazon joins $310m funding for AI startup OdysseyTech in Asia · 2026-06-17
Founders & Team
5- Oliver CameronFounderCo-Founder & CEO
- Jeff HawkeFounderCo-Founder & CTO
- James GrieveVice President of Engineering
- Jessica InmanVice President of Go-To-Market & Operations
- Fabian GüraDistinguished Engineer
Odyssey's about page lists co‑founders Oliver Cameron (Co‑Founder & CEO) and Jeff Hawke (Co‑Founder & CTO) as company leaders; other named leadership includes James Grieve (VP, Engineering), Jessica Inman (VP, Go‑To‑Market & Operations), and Fabian Güra (Distinguished Engineer).
- All people above are taken from the company's About page; the cited company page provides role titles but no extended bios or prior-background lines for these individuals.
- Crunchbase and other aggregators list additional founders (e.g., Jake Abrams, Naman Muley) not named as co‑founders on Odyssey's own About page; those aggregator-only names are not marked as founders here.
- No LinkedIn profile URLs or detailed background statements were present in the cited company pages.
Market Size
No public, dedicated TAM for "world models" (general-purpose multimodal physical-world simulators) was found. The nearest published segment that contains Odyssey's target use cases is the global Model Based Enterprise market (USD 13.6B, 2024, MarketsandMarkets). Because Odyssey publishes only that its pricing is usage-based (no unit or ACV figures) and there are no public ARR/customer counts, a grounded bottom-up SAM cannot be calculated and SOM is not estimated.
- MarketsandMarkets ↗USD 13.6 billion in 2024 · Global · 14.9% CAGR
- IMARC Group ↗USD 10.5 Billion in 2025 · Global · 17.29% CAGR
- Business Research Insights (3D Models Market) ↗USD 2.09 Billion in 2026 · Global · 15.1% CAGR
- Company pricing modelUsage-based (pricing model published on odyssey.ml; no unit prices or tiers published)odyssey.ml (company website, known facts)
- Recent traction / scaleRaised $310M Series B; post-money valuation $1.45B; ~55 employees; offices in Palo Alto, London, Zurich (June 2026)odyssey.ml (company website, known facts)
- Reason SAM not derivedNo published ACV or per-unit usage prices on odyssey.ml and no publicly disclosed customer counts or ARR to ground a bottom-up ACV×count SAM.odyssey.ml (company website, known facts) and public filings/search results (no pricing/ARR disclosed)
- TAM segment choiceUse 'Model Based Enterprise' as the nearest published segment that includes digital-twin/simulation and model-based engineering software/services relevant to world models.MarketsandMarkets (model based enterprise market report)
- There is no published, industry-standard "world models" market TAM; the Model Based Enterprise figure is an adjacent, broader category (digital twins/model-based engineering) and is an imperfect proxy for world-model API demand.
- SAM is null because Odyssey's site discloses a usage-based pricing model but no unit prices, ACV, customer counts or ARR needed for a bottom-up SAM.
- SOM is not estimated because a SOM must be grounded in a bottom-up SAM and/or the company's disclosed revenue/ARR; public materials show strong funding/scale but no revenue figures to convert that scale into a credible near-term market share.
- Many cited reports are commercial market-research products; figures may use different definitions (software vs. services splits are not provided) and some sources are forecasts rather than observed revenue.
Competitors
Top 5Competitive set splits between domain-specific simulators (autonomy, gaming) and research labs; Odyssey sits as a commercially available, general-purpose multimodal world-model leader with strong productization and API distribution.
- World Labs (Marble)marble.worldlabs.ai ↗Creates persistent, editable 3D worlds from images/videos; Odyssey emphasizes dynamic, causal multimodal simulations with long-horizon interactivity, stronger for robotics and control.
- Decartdecart.ai ↗Builds Oasis real-time photorealistic simulators for autonomous-vehicle training; Odyssey offers broader causal multimodal world models, trading domain-specific automotive fidelity for generality.
- AMI Labsamilabs.xyz ↗Yann LeCun's lab building foundational world models; Odyssey already ships Odyssey-2 Pro, giving go-to-market advantage over AMI Labs' research-phase focus.
- Genie 3genie3.net ↗Generates continuous interactive video environments; Odyssey differentiates with causal, multimodal prediction and higher physical-accuracy simulations suitable for long-horizon tasks.
- World Simulator AIworldsimulator.ai ↗Provides AI-driven simulation platforms for testing autonomous systems; Odyssey's strength is multimodal causality and APIs for broader research and robotics use-cases.
Funding
- Seed Round$9MGoogle Ventures leadQasar YounisGuillermo RauchKyle VogtSoleio Soumith ChintalaElad GilBradley HorowitzCrunchbase ↗
- Series A$27 millionMay 28, 2025EQT Ventures leadAir Street CapitalGoogle VenturesGVTechCrunch ↗
- Series B$310 millionJune 17, 2026Natural Capital leadIQTKyle VogtAMD VenturesGuillermo RauchGoogle VenturesSignalRankGarry TanAmazonEQTQasar YounisGVJeff DeanElad GilReuters ↗
Odyssey, a Palo Alto AI lab building general-purpose multimodal world models, has disclosed a $27M Series A (May 28, 2025) and a $310M Series B (June 17, 2026) led by Natural Capital that values the company at $1.45 billion.
- The $310M Series B (June 17, 2026) is reported by multiple reputable outlets (Reuters, TechCrunch and others); coverage lists additional angel and institutional participants across notices — the round-level investor list above is the deduplicated union of named participants in the provided sources.
- The $27M Series A is sourced to TechCrunch (May 28, 2025); press and tracker entries vary in how many earlier rounds they list and in totals (e.g., Tracxn shows differing totals/round counts and Crunchbase's structured block in the corpus lists rounds/investors but omits per-round amounts), but PitchBook's total ($337M) matches the sum of the two disclosed rounds above.
- Several unrelated companies named "Odyssey" appear in the wider coverage (for example, a Fortune item about a $9M seed for a visual-AI startup called Odyssey appears to describe a different company) — rounds above are attributed only where the text clearly refers to the world-models AI lab (Palo Alto, founders Oliver Cameron & Jeff Hawke or explicit world-models description).
- Some aggregator pages (Crunchbase/Tracxn) in the supplied text list additional rounds/figures but do not provide per-round, attributable details in the supplied snippets; those entries were not itemized unless a press source in the corpus explicitly tied an amount/date to this Odyssey.
Momentum
- FundingJune 17, 2026Odyssey raised $310 million in a Series B at a $1.45 billion valuation.Reuters ↗
- PartnershipJune 17, 2026Odyssey announced a strategic partnership naming Amazon Web Services (AWS) its preferred cloud provider.Business Wire ↗
- FundingMay 28, 2025Odyssey closed a $27 million Series A round.TechCrunch ↗
Recent public signals show an accelerating trajectory: Odyssey completed a $310M Series B (June 17, 2026) and concurrently announced a preferred-cloud partnership with AWS, following a $27M Series A in May 2025.
- Many web hits for the name “Odyssey” refer to unrelated films, vehicles, or logistics companies; only items explicitly referencing the AI world-model lab were included.
- Coverage is concentrated around the June 2026 funding announcement and largely reflects press/agency reports and the company’s disclosed funding history; there are few dated public signals about product launches or customer deployments in the provided material.
Customer Sentiment
No clear third‑party customer reviews or ratings attributable to Odyssey (odyssey.ml) were found; public search results mostly return film pages, unrelated apps, and other companies using the name "Odyssey."
Not enough public customer reviews to judge sentiment — common for early-stage or B2B infrastructure labs.
Search-result pages returned by public sites primarily concern films titled "The Odyssey," unrelated consumer apps, a Shopify 3D‑store app, or other companies with the same name, not odyssey.ml or its world‑model API.
App Store listings for apps named "Odyssey" or "Odyssey AI" appear in search snippets but do not clearly link to odyssey.ml or describe a world‑model API; therefore their reviews cannot be reliably attributed to this company.
Because of name collisions across movies, consumer apps, and other businesses, available public review snippets do not provide grounded evidence about Odyssey (odyssey.ml) customers or product experience.
AI Visibility
We ask ChatGPT WITH web search about the company and show its answers plus the sources it cited — the real AI-search experience. The key signal is share of voice: does the company come up when AI is asked about its space, not just when asked by name.
Models asked ChatGPT · web search
Accuracy of what AI said: Accurate — Correctly identifies Odyssey as an AI lab building general-purpose multimodal world models, names the founders (Oliver Cameron and Jeff Hawke), lists core products, and reports the June 2026 Series B, valuation, and Amazon/AWS relationship.
Odyssey is an artificial intelligence (AI) laboratory specializing in the development of general-purpose world models—causal, multimodal systems designed to predict and interact with the world over extended periods. These models have applications across various sectors, including robotics, science, healthcare, education, and gaming. (odyssey.ml)
Founders and Leadership: The company was founded in 2023 by Oliver Cameron and Jeff Hawke. Oliver Cameron serves as the Chief Executive Officer (CEO), bringing experience from his previous role as co-founder and CEO of autonomous vehicle startup Voyage, which was acquired by GM's Cruise. Jeff Hawke, the Chief Technology Officer (CTO), has a background as an engineer at the UK-based self-driving startup Wayve. (techcrunch.com)
Company Size and Locations: As of June 2026, Odyssey employs approximately 46 individuals and operates offices in Palo Alto, California; London, United Kingdom; and Zurich, Switzerland. (linkedin.com)
Notable Developments: In June 2026, Odyssey secured a $310 million Series B funding round, elevating its valuation to $1.45 billion. This round was led by Natural Capital, with participation from Amazon, AMD Ventures, GV, and other investors. Notably, Amazon's involvement has led to Odyssey designating AWS as its preferred cloud provider and optimizing its models to run on AWS’s Trainium chips. (techcrunch.com)
Key Products: Odyssey has developed several advanced world models, including:
- Odyssey-2: A general-purpose world model that significantly advances the physical accuracy of world simulations.
- Starchild-1: A real-time multimodal world model capable of generating both visual and auditory simulations.
- Agora-1: A multi-agent world model that enables multiple participants, whether human or AI, to interact within the same real-time world simulation.
- PROWL: A reinforcement learning-driven framework where agents explore game environments to identify and improve upon failures in world model performance.
These innovations position Odyssey at the forefront of AI research, particularly in the realm of world models that extend beyond traditional language models.
Here is a concise list of notable companies and products in the world-models space:
- Google DeepMind: Developed the Genie series, with Genie 3 generating real-time interactive 3D worlds from text prompts. (en.wikipedia.org)
- NVIDIA: Released Cosmos 3, an advanced open AI world model enhancing robots' and autonomous vehicles' understanding of real-world environments. (axios.com)
- AMI Labs: Founded by Yann LeCun, focusing on world models that learn the structure and dynamics of the physical world. (en.wikipedia.org)
- World Labs: Led by Fei-Fei Li, launched Marble, a world model capable of creating 3D environments from various inputs. (builtin.com)
- Meta Platforms: Developed V-JEPA 2, a self-supervised world model for visual understanding and zero-shot robot control. (world-models.io)
- OpenAI: Known for foundational models like GPT-4, contributing to AI research and deployment. (aiworldnext.com)
- Anthropic: Focuses on building reliable and interpretable AI systems, including world models. (aiworldnext.com)
- Mistral AI: A European company emphasizing open and portable generative AI models. (aiworldnext.com)
- IBM: Researching world models to simulate physical reality for enterprise AI applications. (ibm.com)
- Anduril: Applies AI in defense through platforms like Lattice and autonomous systems. (kiplinger.com)
In the field of artificial intelligence, "world models" refer to systems that enable machines to understand and predict real-world environments, facilitating tasks such as robotics, autonomous driving, and interactive simulations. Several companies are at the forefront of developing and offering products in this space:
- Nvidia: In June 2026, Nvidia unveiled Cosmos 3, an advanced open AI world model designed to enhance the capabilities of robots, autonomous vehicles, and other physical systems. Cosmos 3 is trained on 20 trillion tokens of multimodal data, including images, videos, audio, text, and action data. It is available in two versions: a high-precision "super" model for detailed physics simulations and a lightweight "nano" version for rapid inference. (axios.com)
- World Labs: Founded by Fei-Fei Li, World Labs focuses on spatial intelligence by building foundational world models capable of perceiving, generating, reasoning, and interacting with 3D environments. Their product lineup includes Marble, a tool for creating high-fidelity 3D worlds from various inputs; the World API, which allows developers to generate explorable 3D worlds programmatically; and Spark, a web-native renderer for dynamic 3D scenes. (allyticstechperspectives.com)
- Google DeepMind: In 2024, DeepMind introduced Genie, a model that learned interactive environments from unlabeled internet videos. By 2025, Genie 3 was capable of producing photorealistic, real-time interactive worlds from text prompts, supporting persistent 3D environments and real-time interaction. Waymo adopted Genie 3 in 2026 to create specialized world models for autonomous driving simulations. (en.wikipedia.org)
These companies are leading the development of world models, contributing significantly to advancements in AI's ability to understand and interact with complex real-world environments.
Odyssey is not surfaced unprompted in the category or recommendation lists, but when asked by name the assistant produced an accurate, up-to-date description that matches public reporting on founders, products, funding, and the AWS partnership. The company's omission from the unprompted lists suggests low spontaneous visibility in the sampled assistant outputs despite reliable direct identifiability.
- AI answers can vary run-to-run and across prompt phrasings; a single snapshot may not represent all assistant behavior.
- Omission from category/recommendation answers may reflect list curation, recency, or source selection rather than company importance.
- Public reporting on young AI labs can change quickly; confirm critical facts (funding, valuation, leadership) from primary press coverage or filings before investment decisions.
Risks & Red Flags
No material public risk signals found. An absence of public concerns isn’t a clean bill of health — early or private companies may simply not generate coverage.
This assessment is limited to independently verifiable, publicly-available third‑party sources (news outlets, regulators, court records, trade press). It does not include non-public information, proprietary databases not accessible here, or private disputes that have not been reported publicly.
An absence of public risk signals is not a clean bill of health: early-stage or privately-held technology companies often have limited public reporting, so material issues could exist but remain undisclosed.
Search coverage is current through June 2026 (the date of the company's disclosed Series B).