Inside Decart’s $300M Funding Round and Its Bet on Real‑Time AI World Models

The round values Decart at roughly $4 billion, a jump from the company’s earlier $3.1 billion valuation after a $100 million raise the previous year.

For investors, the thesis is clear: if AI systems move from static outputs to continuous, interactive environments, the companies that build the infrastructure for those systems could become a new foundational layer of the AI stack.

Strategic Backers: Nvidia and Amazon

Two of the most notable participants in Decart’s ecosystem are Nvidia and Amazon Web Services.

Nvidia joined the funding round as an investor, reflecting the tight connection between advanced AI models and high‑performance compute hardware.

Amazon’s role appears somewhat different. Reports describe the company as joining as a strategic customer, while AWS also collaborates with Decart on infrastructure for real‑time AI systems.

Part of that relationship includes optimizing Decart’s Lucy model for AWS Trainium accelerators, Amazon’s custom chips for machine‑learning workloads.

Taken together, these partnerships hint at a key part of Decart’s strategy: building AI systems that can run across multiple accelerator ecosystems rather than relying exclusively on Nvidia GPUs.

The Product Stack: DOS, Lucy, and Oasis

Decart’s platform consists of three main components designed to work together.

DOS: The Infrastructure Layer

DOS is Decart’s core systems platform—an optimized inference and training stack designed for low‑latency AI systems.

The goal is to make AI models faster and more efficient so they can operate continuously rather than responding to one prompt at a time.

Lucy: Real‑Time World Generation

Lucy is a world‑model system focused on immersive environments and real‑time video generation.

The company describes Lucy as capable of generating “real‑time, infinite video” that evolves continuously rather than producing isolated clips.

Oasis: World Models for Physical AI

Oasis targets simulations and environments for robotics and other physical‑world systems.

According to Decart, Oasis can generate interactive environments that respond in real time, which could be used to train robotics systems or simulate physical environments.

Earlier versions of the company’s technology also focused on reducing GPU costs and enabling more natural interaction for AI agents, suggesting that efficiency and real‑time performance are central design goals.

AWS Trainium and a Cross‑Chip Strategy

One of the more interesting signals from Decart’s partnerships is its use of AWS Trainium accelerators.

The company is working with AWS to optimize Lucy for Trainium3, Amazon’s AI‑specific chip designed to compete with traditional GPU‑based training and inference.

This matters for two reasons:

• It shows Decart designing its infrastructure to run on multiple chip architectures.
• It aligns the company with both GPU ecosystems (like Nvidia) and custom AI silicon platforms (like Trainium).

Decart itself describes its technology stack as spanning “the whole computational stack—from hardware to world models.”

However, publicly available reports do not provide detailed performance benchmarks or cost comparisons for Lucy running on Trainium versus GPUs.

The Big Vision: Real‑Time World Models

At the center of Decart’s strategy is a concept known as world models—AI systems capable of generating and simulating environments that evolve continuously over time.

Instead of producing static outputs like text or images, these systems aim to create persistent, interactive digital worlds that respond instantly to users or machines.

Potential applications include:

• Gaming: dynamically generated worlds and gameplay environments
• Media and content creation: real‑time interactive video and storytelling
• Robotics: simulated environments for training autonomous systems
• Physical AI: modeling real‑world interactions for drones, vehicles, or manufacturing

The company positions its technology as enabling AI that can “run instantly, continuously, and efficiently.”

What the Funding Round Really Signals

The significance of Decart’s $300 million raise goes beyond capital.

It reflects a growing belief among investors and technology companies that AI’s next phase may revolve around real‑time, interactive systems rather than static outputs.

The structure of the round highlights three strategic themes:

• Infrastructure matters: Decart is building systems software (DOS) alongside models.
• Hardware alignment: partnerships with Nvidia and AWS position it across competing compute ecosystems.
• World‑model ambition: Lucy and Oasis aim to generate persistent digital environments for both virtual and physical applications.

What remains uncertain is execution. Public reporting confirms the company’s funding, partnerships, and technical direction, but independent benchmarks, large‑scale deployments, and revenue details remain limited.

Still, the scale of the round—and the caliber of investors involved—suggests the industry sees real‑time AI worlds as a major frontier in the next wave of artificial intelligence.