Decart: AI Video Startup Eyed in $7B Acquisition

Decart, a fast-rising startup specializing in real-time generative video, has reportedly emerged as a roughly $7 billion acquisition target—a valuation that underscores how central real-time synthetic media has become to the broader AI landscape. For a company founded only recently, the figure is a striking signal of investor and strategic appetite for the technology that powers live AI video transformation.

Who Is Decart?

Decart is an AI research and product company that has built a reputation around real-time generative world models—systems capable of generating and transforming video frames on the fly rather than rendering output in slow, offline batches. Unlike traditional text-to-video pipelines that take seconds or minutes to produce a clip, Decart's approach emphasizes interactivity and low latency, a technically demanding requirement that pushes the limits of model efficiency and inference optimization.

The company drew widespread attention with Oasis, an interactive, AI-generated game environment that produced playable frames in real time without a conventional game engine—every frame hallucinated by a neural network responding to user input. That demo was more than a novelty: it demonstrated that a generative model could maintain temporal and spatial coherence while responding to live control signals, one of the hardest problems in synthetic video.

Decart followed with real-time video transformation tools (marketed under names such as Mirage and Lucy), which allow live video streams to be restyled, re-skinned, or transformed into entirely different visual worlds as they happen. This is the kind of capability that sits squarely at the intersection of creative tooling and synthetic media—and, inevitably, at the center of digital authenticity concerns.

Why a $7B Valuation Matters

For context, real-time generative video is one of the most technically expensive problems in AI. Producing coherent frames at interactive frame rates requires aggressive model distillation, custom inference stacks, and often bespoke hardware optimization. A $7B acquisition interest suggests that larger players view this capability as strategically scarce—something easier to buy than to build from scratch.

The deal talk places Decart in the same conversation as other high-value generative media players like Runway, Pika, and the video divisions of larger labs. Where most of those competitors still operate in a generate-then-review workflow, Decart's real-time angle differentiates it. Real-time generation unlocks use cases that batch models can't easily serve: live streaming filters, interactive entertainment, virtual production, gaming, and immersive experiences where latency is the entire product.

The Technical Challenge Behind Real-Time Video

What makes Decart's work notable is the engineering required to collapse the generation loop. Diffusion-based video models typically rely on many denoising steps per frame, which is fundamentally incompatible with interactive frame rates. Achieving real-time output generally demands techniques like step distillation, consistency models, or autoregressive frame prediction optimized for streaming. Maintaining temporal consistency—so objects and scenes don't flicker or warp between frames—is a persistent technical hurdle that separates impressive demos from production-grade systems.

If Decart has genuinely solved real-time coherence at scale, that intellectual property is precisely the kind of asset that commands a premium acquisition price. It represents not just a product, but a defensible technical moat in a crowded generative video field.

Implications for Digital Authenticity

Real-time video transformation is a double-edged capability. The same technology that enables live virtual production and interactive entertainment also lowers the barrier to live deepfakes—synthetic video that can be generated and streamed without the delay that currently makes real-time impersonation difficult. As these tools mature and get absorbed into larger platforms via acquisitions like this one, the pressure on detection and provenance systems intensifies.

Content authentication frameworks, watermarking standards, and provenance tracking will need to keep pace with generation technology that operates at the speed of a live stream. A $7B consolidation event is a reminder that the capability side of this arms race is extremely well-funded—and that authenticity infrastructure must scale accordingly.

The Bigger Picture

Whether or not the reported acquisition closes at $7B, the figure itself is the story. It marks real-time generative video as a strategically critical frontier, not a research curiosity. For observers of synthetic media, Decart's trajectory from startup to multi-billion-dollar target is a concrete data point on how quickly value is accruing in the AI video space—and how the next generation of synthetic content may arrive in real time.

Stay informed on AI video and digital authenticity. Follow Skrew AI News.