ComfyUI raises $30M at $500M valuation as open-source AI creative platform hits Hollywood studios

ComfyUI raises $30M at $500M valuation

ComfyUI, the open-source node-based AI workflow platform, has raised $30M led by Craft Ventures at a $500M valuation. Founder and CEO Yoland Yan says the company's pitch to professional studios is control — granular, deterministic control over generative AI outputs that text-prompt tools like Midjourney can't offer.

The platform lets users adjust camera angles, facial features, posture, and model behavior through a node-based interface, and run variations in parallel rather than prompting sequentially. Yan says recent Coca-Cola and Super Bowl ads were produced using ComfyUI, and that top studios are already using it in pre-production — feeding green-screen footage through the system to get full 3D CG renders in roughly five minutes.

“We raised $30,000,000. The round was led by Craft Ventures. We are here to destroy our enemies and make open source creative tooling win. Comfy is a node-based system that aims to give the highest leverage of control and quality for content creation — the recent Coca-Cola ad and Super Bowl ads were all made using Comfy.”

Comfy Cloud, launched about five months ago, is the commercial layer. Studios that previously ran ComfyUI locally on their own hardware can instead tap ComfyUI's compute infrastructure. Yan describes stitching that infrastructure together for professional creative clients as genuinely hard, which is where the monetization lives.

On the question of whether ComfyUI should train its own models — the Cognition path — rather than stay at the application layer, Yan argues the visual creative market is structurally different from coding. Even if a foundation model could generate a feature film tomorrow, professional creatives would still want iterative, taste-driven control over the output. Comfy is where that control lives, model-agnostic, supporting both open and closed-source foundation models including CogVideoX.

Yan's read on long-form AI video is that the bottleneck right now isn't quality — it's model velocity. By the time a team finishes a feature-length piece, the underlying model has shifted enough that earlier scenes need to be rerun. Studios therefore focus on shorter clips and use ComfyUI to accelerate iteration, not replace production.

The clearest near-term commercial argument Yan makes to AI-skeptical Hollywood clients is that ComfyUI keeps humans in the loop. It automates the tedious parts — colorization, upscaling, downscaling — without displacing creative judgment. He points to a measurable proxy: job postings for ComfyUI artists and specialists are growing across hiring platforms.

Yan frames the competitive target plainly — open-source creative tooling winning against proprietary incumbents, with Adobe the obvious implied reference.