ComfyUI just closed a $30 million round at a $500 million valuation, according to TechCrunch AI. Craft Ventures led the deal, with Pace Capital, Chemistry, and TruArrow joining in. The startup builds node-based tools that let creators take granular control over image, video, and audio outputs from diffusion models, and the new round more than doubles where it stood after its $19 million Series A in late 2024.
This is significant because it’s a bet on a specific thesis: prompt boxes alone aren’t enough for serious creative work, no matter how good the foundation models get.
What ComfyUI actually does
ComfyUI started as an open-source project in 2023, right when diffusion models were still spitting out hands with six fingers. Instead of typing a prompt and praying, you wire up a graph of nodes that controls each step of the generation pipeline. You can swap models, lock down specific regions, chain refiners, and keep the parts that already work while changing only what you want to change.
That workflow grew a cult following among technical artists. The startup now claims over 4 million users across visual effects, animation, advertising, and industrial design. TechCrunch AI notes that “ComfyUI artist” and “ComfyUI engineer” have started showing up as actual job titles on studio listings.
Why the funding makes sense now
Co-founder and CEO Yoland Yan framed the problem in plain terms to TechCrunch AI. “If you think about your typical prompt-based solution, like Midjourney or ChatGPT, you ask for something, it [gets only] 60% to 80% there,” he said. “But to change that remaining 20%, you have to try this slot machine.”
The slot machine metaphor is the whole pitch. Tweak a prompt to fix one detail and the model often rewrites the parts you already loved. Yan’s argument is that node-based control turns that gamble into actual engineering.
“In the world where AI slop is going to be everywhere, the Comfy version of human-in-the-loop approach is going to win out most of the eyeballs in the end,” he told TechCrunch AI.
How this compares to what came before
A year ago, the conventional wisdom was that better foundation models would absorb the need for tools like ComfyUI. Sora, Veo, Midjourney V7, and the latest image stacks all got dramatically sharper. The hands fixed themselves.
What didn’t fix itself: precision. Pros doing client work need pixel-level control, consistent characters across shots, tight masking, and reproducible pipelines. The gap between “impressive demo” and “shippable asset” is exactly where ComfyUI lives, and investors clearly think that gap isn’t closing.
For context, the competitive landscape is thinning. Weavy, a similar player, was acquired by Figma last year. That leaves ComfyUI as the dominant standalone option for the open, modular workflow camp.
What this means for practitioners
A few takeaways worth chewing on:
- Node-based workflows aren’t a niche anymore. If “ComfyUI engineer” is a real job title, that skill set is becoming part of the production pipeline alongside compositing and 3D.
- Prompting is the floor, not the ceiling. Teams shipping AI-generated media at scale are moving toward graph-based control because prompt-only output costs them too many regenerations.
- Open-source roots still pay off. ComfyUI’s traction came from being free and hackable first. The commercial layer is being built on top of an already-loved tool, not sold into a cold market.
- Expect more verticalized tooling. With $30 million in fresh capital, ComfyUI can push into advertising, VFX studios, and industrial design with dedicated features.
What’s next
The round gives ComfyUI runway to expand its commercial offering while foundation models keep improving underneath it. The bet is that the better the models get, the more valuable precise control over them becomes. If Yan is right about “AI slop” flooding feeds, the tools that let humans steer the output will be the ones professionals reach for.
More details on the round and ComfyUI’s roadmap are in the original report at TechCrunch AI.