Real-World Telemetry World Foundation Models AI Frontier Labs

Value Proposition & Features

Decart is a frontier AI lab founded in 2023 that builds real-time world models and infrastructure for running generative AI with very low latency. [^E1] Its stated product family includes Oasis, Lucy, and the Decart Optimization Stack (DOS). [^E1]

• Real-time world models: Generates responsive interactive environments for applications such as robotics, autonomous-driving development, and digital-world simulation. [^i7scjj]
• Lucy: Real-time video generation and transformation, including live video editing. [^i7scjj]
• Oasis: Simulated environments intended for training and testing robotics and autonomous-driving systems. [^i7scjj]
• Decart Optimization Stack (DOS): Infrastructure for optimizing AI workloads across hardware and inference systems. [^hp6bls]
• Hardware portability: DOS is described as supporting NVIDIA, AWS Trainium, and Google TPU platforms. [^hp6bls]
• Low-latency inference: Decart positions its infrastructure around faster, more efficient model execution and millisecond-latency experiences. [^E1]
• Generative interactive models: Combines model development with infrastructure for responsive visual and physical-world applications. [^E1]

Product Roadmap / Announcements

As of October 12, 2026,

• July 2026: Lucy 2.5 was announced as an improved real-time video model with higher-quality transformations. [^hp6bls]
• May 2026: Decart announced a $300 million financing round to expand low-latency AI infrastructure and future Lucy and Oasis models. [^hp6bls]

Recent Developments

• October 2026: Amazon was reportedly discussing an acquisition of Decart at a valuation of approximately $7 billion; the negotiations were described as unfinalized. [^6rtzw8]
• September 2026: Anthropic reportedly abandoned a potential acquisition after due diligence, following discussions valuing Decart at roughly $6 billion. [^i7scjj]
• May 2026: Decart raised $300 million in a Radical Ventures-led round, with NVIDIA participating. [^be3f3z]

History and Origin Story

Decart was founded in 2023 by Dean Leitersdorf, Orian Leitersdorf, and Moshe Shalev, according to multiple reports. [^qy3tdp] [^6rtzw8] The company developed from real-time generative video and world-model research into a broader platform spanning interactive models and AI optimization infrastructure. [^4ejmql] [^hp6bls]

Fundraising History

The supplied company material reports total funding of $456 million, including the rounds listed above. [^E1] Contemporary reporting describes Decart as having raised more than $400 million and being valued at nearly $4 billion after the May 2026 round. [^xqc7cv]

Atreides Management Benchmark NVIDIA Radical Ventures Sequoia Capital Valor Equity Partners Zeev Ventures Adobe Ventures [^xqc7cv]

Notable Team Members

Dean Leitersdorf is a co-founder of Decart and is identified in reporting alongside Orian Leitersdorf and Moshe Shalev. [^6rtzw8]

Orian Leitersdorf is a co-founder of Decart and the brother of Dean Leitersdorf, according to reporting on the company’s founding team. [^6rtzw8]

Moshe Shalev is a co-founder of Decart and part of the founding team established in 2023. [^qy3tdp]

Market Sizing

Category, Market Size, and Category Growth

Decart operates across several overlapping categories: generative AI infrastructure, real-time video generation, world models, and physical-AI simulation. [^E1] [^i7scjj] No reliable analyst-firm market-size or category-growth estimate specific to Decart’s combined market was found in the available results.

Competitive Landscape

Who it's for, who it's not for

Decart is aimed at AI developers, robotics and autonomous-driving teams, model operators, and organizations that need responsive generative environments or efficient inference across heterogeneous hardware. [^hp6bls] [^i7scjj]

It is not positioned as a general-purpose consumer chatbot, conventional video-editing suite, or ordinary cloud infrastructure provider; its products focus on model execution, real-time visual generation, and interactive-world simulation. [^E1] [^4ejmql]

Viable Alternatives

• NVIDIA Omniverse: A plausible alternative for teams prioritizing physically based 3D simulation and robotics workflows.
• World Labs: An alternative for generative 3D-world construction and spatial intelligence.
• Runway: An alternative for generative video creation and transformation, though not necessarily equivalent in world-model infrastructure.
• Google DeepMind: A potential alternative for organizations seeking frontier world-model and robotics research.
• AWS AI infrastructure: A potential alternative for model-serving and hardware-optimization requirements, especially where cloud-native deployment is the priority.

Competitor Table

---

Sources

¹: ☕ Nvidia looking to acquire Reflection AI & Sony sells 419 ... [^4ejmql]: Amazon is reportedly circling Decart after Nvidia and ... ³: Sources: Amazon is in talks to buy AI startup Decart in a ... [^xqc7cv]: Amazon in talks to acquire AI startup Decart in $7 billion deal, report says ⁵: Gaming Startup Funding Levels Up A Bit In 2026 [^hp6bls]: www.discords.ai · wiki · decart-video-simulation-tech-ai-gameDecart AI Explained: $300M Funding, Lucy, Oasis & Gaming [^qy3tdp]: AI News, 12 Oct 2026: Cloudflare Buys Deno, Manus $500M [^6rtzw8]: Amazon in talks to buy Israel's Decart at $7b valuation - report [^i7scjj]: Amazon in talks to buy AI startup Decart for around $7 ... [^be3f3z]: Amazon in talks to buy AI startup Decart for around $7 ...

Sources (retrieved)

[^E1]: Exa.ai API response for data on Decart