Fal raises $140M Series D led by Sequoia as image editing surpasses AI video in revenue

Fal, the AI infrastructure platform, closed a $140 million Series D led by Sequoia Capital, with meaningful participation from Kleiner Perkins and Nvidia. The round is the company's third fundraise in 2025 alone, signaling sustained investor appetite for inference and model-serving infrastructure.

Image editing has overtaken AI video as Fal's largest revenue driver, a result that surprised the company's own leadership. Fal entered 2025 expecting AI video to carry growth, and while that segment performed strongly, image editing emerged as the dominant category after the first compelling editing models arrived around May 2025. Customers generating on the platform span major advertisers, retail platforms, design and productivity apps, and movie studios.

“Yurtseven: 'Today, we announced our $140M Series D led by Sequoia with meaningful participation from Kleiner Perkins and Nvidia. What was surprising to us is the rise of image editing — one of the first good models came out around May, and now it's a bigger part of our business than AI video. That was surprising to us.'”

The generative media model market is structurally more fragmented than the LLM space. The leading image model shifts frequently, with Google's Imagen 3 (internally code-named Nana Banana, using a smaller Gemini model as the underlying architecture) currently holding the top position for image editing. Open-source communities and Chinese labs have consistently closed gaps within months of frontier releases, partly because seeing market demand clarifies what to build, and partly because research techniques propagate quickly.

AI video remains pre-mainstream on the commercial side. Producing a high-quality AI video ad currently requires more effort than a conventional shoot, limiting adoption to technically proficient creators. Cling, developed by a Chinese lab, is identified as one of the stronger current video models, with recently announced editing capabilities drawing attention. The critical missing feature across video models is character and scene-to-scene consistency, viewed as the unlock for broad commercial adoption in 2026.

Fal's go-to-market thesis is that the image editing trajectory will repeat in video as model quality and usability converge, pulling in studios and retailers at scale. The company is positioning its platform to capture that volume regardless of which underlying model leads at any given moment.